0% found this document useful (0 votes)

710 views

SAS SQL - Course Notes PDF

Uploaded by

augustocgn

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

710 views

SAS SQL - Course Notes PDF

Uploaded by

augustocgn

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 366

SAS® SQL 1: Essentials

Course Notes
SAS® SQL 1: Essentials Course Notes was developed by Peter Styliadis, Charlot Bennett, Johnny
Johnson, and Mark Jordan. Additional contributions were made by Michele Austin, Brittany Coleman,
Bruce Dawless, Davetta Dunlap, Marty Hultgren, John McCall, Rich Papel, Ross Richards, Lorilyn
Russell, Allison Saito, Ian Sedgwick, Charu Shankar, Jim Simon, Theresa Stemler, Stacey Syphus,
Chris Warters, and Anna Yarbrough. Instructional design, editing, and production support was
provided by the Learning Design and Development team.

SAS and all other SAS Institute Inc. product or service names are registered trademarks or
trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies.

SAS® SQL 1: Essentials Course Notes

Copyright © 2019 SAS Institute Inc. Cary, NC, USA. All rights reserved. Printed in the United States
of America. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mec hanical, photocopying, or otherwise,
without the prior written permission of the publisher, SAS Institute Inc.

Book code E71409, course code LWSQ1M6/SQ194, prepared date 11Jun2019. LWSQ1M6_001

ISBN 978-1-64295-094-6
For Your Infor mation iii

Table of Contents

Lesson 1 Essentials ...........................................................................................1-1

1.1 Setting Up for the Course ..................................................................................1-3

1.2 What Is SQL? ................................................................................................ 1-10

1.3 Introduction to the SQL Procedure ................................................................... 1-13

Demonstration: Exploring the customer Table .............................................. 1-21

1.4 Solutions ....................................................................................................... 1-26

Solutions to Activities and Questions .......................................................... 1-26

Lesson 2 PROC SQL Fundamentals ...................................................................2-1

2.1 Generating Simple Reports ...............................................................................2-3

Demonstration: Creating Simple Reports ..................................................... 2-17

Demonstration: Assigning Values Conditionally ............................................ 2-24

Practice ................................................................................................... 2-27

2.2 Summarizing and Grouping Data ..................................................................... 2-31

Demonstration: Summary Functions ........................................................... 2-35

Demonstration: Analyzing Groups of Data ................................................... 2-41

Demonstration: Summarizing Data Using a Boolean ..................................... 2-47

Practice ................................................................................................... 2-50

2.3 Creating and Managing Tables ........................................................................ 2-53

2.4 Using DICTIONARY Tables ............................................................................. 2-63

Demonstration: Using DICTIONARY Tables ................................................. 2-66

Practice ................................................................................................... 2-69

2.5 Solutions ....................................................................................................... 2-71

Solutions to Practices................................................................................ 2-71

Solutions to Activities and Questions .......................................................... 2-75

iv For Your Information

Lesson 3 SQL Joins ...........................................................................................3-1

3.1 Introduction to SQL Joins ..................................................................................3-3

3.2 Inner Joins .......................................................................................................3-8

Demonstration: Performing an Inner Join with PROC SQL ............................ 3-10

Demonstration: Performing an Inner Join with Four Tables ............................ 3-16

Practice ................................................................................................... 3-23

3.3 Outer Joins .................................................................................................... 3-28

Demonstration: Performing a Full Join with PROC SQL ................................ 3-35

Practice ................................................................................................... 3-40

3.4 Complex Joins ............................................................................................... 3-45

Demonstration: Performing a Reflexive Join................................................. 3-47

3.5 Solutions ....................................................................................................... 3-54

Solutions to Practices................................................................................ 3-54

Solutions to Activities and Questions .......................................................... 3-60

Lesson 4 Subqueries..........................................................................................4-1

4.1 Subquery in WHERE and HAVING Clauses ........................................................4-3

Demonstration: Subquery That Returns a Single Value ...................................4-7

Demonstration: Subquery That Returns Multiple Values ................................ 4-12

Practice ................................................................................................... 4-19

4.2 In-Line Views (Query in the FROM Clause)....................................................... 4-25

Demonstration: Using an In-Line View ......................................................... 4-29

Practice ................................................................................................... 4-39

4.3 Subquery in the SELECT Clause ..................................................................... 4-43

Demonstration: Remerging Summary Statistics............................................ 4-46

Practice ................................................................................................... 4-51

4.4 Solutions ....................................................................................................... 4-54

Solutions to Practices................................................................................ 4-54

Solutions to Activities and Questions .......................................................... 4-61

For Your Infor mation v

Lesson 5 Set Operators ......................................................................................5-1

5.1 Introduction to Set Operators .............................................................................5-3

5.2 INTERSECT, E XCEPT, and UNION ...................................................................5-9

Demonstration: Using the UNION Set Operator............................................ 5-18

Practice ................................................................................................... 5-24

5.3 OUTER UNION .............................................................................................. 5-28

Demonstration: Using the OUTER UNION Set Operator ............................... 5-30

Practice ................................................................................................... 5-34

5.4 Solutions ....................................................................................................... 5-36

Solutions to Practices................................................................................ 5-36

Solutions to Activities and Questions .......................................................... 5-39

Lesson 6 Using and Creating Macro Variables in SQL ........................................6-1

6.1 Creating User-Defined Macro Variables ..............................................................6-3

6.2 Creating Data-Driven Macro Variables with PROC SQL .......................................6-6

Demonstration: Using a PROC SQL Data-Driven Macro Variable.................. 6-10

Demonstration: Concatenating Values in Macro Variables ............................. 6-17

Practice ................................................................................................... 6-22

6.3 Solutions ....................................................................................................... 6-28

Solutions to Practices................................................................................ 6-28

Solutions to Activities and Questions .......................................................... 6-33

Lesson 7 Acce ssing DBMS Data with SAS/ACCESS ® .........................................7-1

7.1 Overview of SAS/ACCESS Technology ..............................................................7-3

7.2 SQL Pass-Through Facility ................................................................................7-6

Demonstration: Using an SQL Pass-Through Query ..................................... 7-13

7.3 SAS/ACCESS LIBNAME Statement................................................................. 7-18

Demonstration: Using the SAS/ACCESS LIBNAME Statement...................... 7-20

vi For Your Information

7.4 FEDSQL Procedure ........................................................................................ 7-27

7.5 Solutions ....................................................................................................... 7-39

Solutions to Activities and Questions .......................................................... 7-39

For Your Infor mation vii

To learn more…
For information about other courses in the curriculum, contact the
SAS Education Division at 1-800-333-7660, or send e-mail to
[email protected]. You can also find this information on the web at
http://support.sas.com/training/ as well as in the Training Course
Catalog.

For a list of SAS books (including e-books) that relate to the topics
covered in this course notes, visit https://www.sas.com/sas/books.html or
call 1-800-727-0025. US customers receive free shipping to US
addresses.
viii For Your Information
Lesson 1 Essentials
1.1 Setting Up for the Course ............................................................................................ 1-3

1.2 What Is SQL? ............................................................................................................. 1-10

1.3 Introduction to the SQL Procedure ............................................................................ 1-13

Demonstration: Exploring the customer Table ........................................................... 1-21

1.4 Solutions ................................................................................................................... 1-26

Solutions to Activities and Questions........................................................................ 1-26
1-2 Lesson 1 Essentials

1.1 Setting Up for the Course

Course Overview

Fundamentals Joins

PROC
FEDSQL

Subqueries
Accessing DBMS Data SQL

Using and Creating

Macro Variables in SQL Set Operators
3
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Course Overview
Demonstration Performed by your instructor as an example for you
to observe
Activity Short practice opportunities for you to perform in SAS,
either independently or with the guidance of your
instructor
Practice Extended practice opportunities for you to work on
independently
Case Study A comprehensive practice opportunity at the end
of the class

4
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Choosing a Practice Level

Level 1 Solve basic problems with step-by-step guidance

Level 2 Solve intermediate problems with defined goals
Challenge Solve complex problems independently with SAS Help
and documentation resources
Choose one
practice to do in
class based on your
interest and skill
level.

5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When you come to a practice, you can choose an appropriate level: basic, intermediate, or complex.

SAS Programming Interfaces

SAS Windowing
SAS Studio Environment

SAS Enterprise Guide

6
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SAS Programming Windows

write
view
and view
messages
submit 31
32
p ro c s ql ;
c re a te t a bl e t a rg e tc u st as
results
33 s el e ct * from SAS
code 34 f ro m s q. c us t om e r
35 w he re ba n ki d =. ;
pr oc s ql ; NO T E: Ta b le WO R K. TA R GE T CU S T c re a te d , w it h
49 5 4 r ow s a n d 2 2 co l um n s.
cr ea te t ab le t a rg et cu st a s
36 q ui t ;
se le ct * NO T E: PR O CE D UR E S QL us e d ( To t al pr o ce s s
ti m e) :
f ro m sq .c us t om er re a l t im e 0. 1 1 s ec o nd s
w he re b an ki d =. ; cp u t i me 0. 0 3 s ec o nd s

qu it ;
37
38 p ro c s ql ;
pr oc s ql ; 39 s el e ct *
40 f ro m w or k .t a rg e tc u st ;
se le ct * 41 q ui t ;
NO T E: PR O CE D UR E S QL us e d ( To t al pr o ce s s
f ro m wo rk .t a rg et cu st ; ti m e) :
re a l t im e 2. 4 8 s ec o nd s
qu it ; cp u t i me 2. 4 3 s ec o nd s

Results and
Editor Log Output Data
7
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Accessing the Course Files

Make a note of
the location of
course
your course
files
activities files folder.

data

database

demos

practices
8
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For this course, you use a variety of data files and SAS programs. The SAS program files are
organized into folders for activities, demos, and practices.

Accessing the Course Files

Programs in the
activities, demos,
course and practices
files folders follow this
activities
naming convention.

data

database

demos
s104d01.sas
practices SQL, Lesson 4, demo 1
9
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Creating the Course Data

course
files
activities

data
cre8data.sas
database

demos

practices
10
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Open and run the cre8data.sas program in the data folder to create the tables for the course. If your
files will not be in s:/workshop, change the value of PATH= in the %LET statement to reflect your
actual course folder location.

1.01 Activity (Required)

1. Na vi ga te to the l ocation of the course files.
SAS Studio: In the Na vi gation pa ne, expand Files and Folders.
SAS Enterprise Guide: In the Servers l ist, expand Servers  Local 
Files.
2. Doubl e-click the cre8data.sas fi l e to open the program.
3. Fi nd the %LET s ta tement. As di rected by your i nstructor, provi de the
pa th to your cours e fi les.
4. Run the progra m a nd veri fy tha t a report i s crea ted listing 27 generated
ta bl es.

11
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Sample of Data Used in This Course

customer globalpop
transaction

merchant statecode

bank statepopulation globalfindex

13
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

United States population data was obtained from the US Census Bureau:
• United States Census Bureau: US Census State Population Totals and Components of Change:
https://www.census.gov/newsroom/press-kits/2018/pop-estimates-national-state.html
Global data was obtained from the World Bank:
• The World Bank: Global Financial Inclusion (Global Findex) Database:
https://datacatalog.worldbank.org/dataset/global-financial-inclusion-global-findex-database

• The World Bank: Population estimates and projections:

https://databank.worldbank.org/data/source/population-estimates-and-
projections/Type/TABLE/preview/on

Connecting to a Library to Read SAS Files (Review)

access LIBNAME libref engine "path" ;

data
name of location of
type of data
library data

• eight-character maximum • Engines such as Oracle,

• starts with a letter or Teradata, and Postgres
underscore enable you to read and
• continues with letters, write to other types of
numbers, or underscores data.

14
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the LIBNAME statement, see "LIBNAME Statement" in the SAS® 9.4 and
SAS® Viya® 3.4 Programming documentation. You can also find the direct link in the Course Links
section on the Extended Learning page.

Connecting to a Library to Read SAS Files (Review)

libname sq base "s:/workshop/data";

connect
to the
library SAS tables data located in
library name
(default engine) s:/workshop/data

LIBNAME is a global
statement and does
not require a RUN
statement.

15
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When reading SAS tables, the BASE engine is optional.

1.02 Activity
1. Open libname.sas from the ma i n fol der. Compl ete the LIBNAME s ta tement
to crea te a l i bra ry na med sq tha t rea ds SAS ta bles i n the data fol der. The
path should be the folder where your course files are located.
Note: In Enterpri se Guide, cl ick Libraries  Refresh to upda te the library l ist.
libname sq "s:/workshop/data";

2. Run the code a nd veri fy tha t the l ibrary wa s s uccessfully a ssigned i n the
l og.
3. Sa ve the upda ted libname.sas progra m.
4. Na vi ga te to your l i st of l ibra ries a nd expand the sq l i bra ry. Veri fy tha t the
l i brary exi sts and tables a re a vailable.
16
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Extending Your Learning

Use your Extended

Learning page (ELP) to
download course files and
access additional videos,
papers, and other helpful
resources.

18
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Throughout each lesson, you will see references to topics in the SAS documentation. All links are
provided on the ELP for each lesson.

1.2 What Is SQL?

Tables

Column, variable, or field

customer
CustomerID
FirstName
CustomerID FirstName LastName …
LastName
Row, observation, …
Employed
or record …
Address
…
City
State
…

20
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Collection of Related Tables

customer transaction merchant
CustomerID TransactionID MerchantID

FirstName CustomerID CompanyName

LastName Date Address

Employed MerchantID City

Address Amount State

City BankID …

State
bank
…
BankID
BankName
Address
City
State
…
21
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Structured Query Language (SQL)

Structured Query Language

(SQL) is a standardized,
widely used language
designed to query and
manage data in relational
databases.
SQL

22
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Timeline
Relational Database
Model Proposed

First ANSI Standard First major revision to the Last major revision to the
SEQUEL SQL (SQL-87) ANSI Standard (SQL-2) ANSI Standard (SQL-3)
Developed

SAS PROC SQL

1970 1973 1986 1992 1993 1999

23
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

1970 - Dr. Edgar F. Codd proposes a relational data model (“A Relational Model of Data for Large
Shared Data Banks”)
1973 - Donald D. Chamberlin and Raymond F. Boyce develop SEQUEL (Structured English Query
Language) at IBM
1986 - First ANSI standard for SQL (SQL-87)
1992 - First major revision to the ANSI standard (SQL-2)

1993 - SAS SQL procedure added (based on SQL-2 standard). Although PROC SQL incorporated
many of the standards, SAS SQL is not ANSI compliant.
1999 - Last major revision to the ANSI standard (SQL-3)

SQL Implementations

Any ma ny
more!

Database Management Systems (DBMS) offer different

implementations of SQL.

24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-13

1.3 Introduction to the SQL Procedure

SQL Procedure
SQL SAS
PROC PROC SQL enables the use
of SQL in SAS and includes
non-ANSI-compliant
SAS enhancements.
SQL
SQL

26
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

PROC SQL follows most of the guidelines set by the American National Standards Institute (ANSI) in
its implementation of SQL. However, it is not fully compliant with the current ANSI standard for SQL.
For more information about SAS and the ANSI standard, see "PROC SQL and the ANSI Standard"
in the SAS SQL Procedure User’s Guide documentation. You can also find the direct link in the
Course Links section on the ELP.

SAS Programming Process

Analyze and
Access Explore Prepare Export
report on
data data data results
data

PROC SQL

27
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

As you go through the process of making data meaningful and actionable using SQL, you will likely
follow these basic steps: access data or read it with a program, explore the data to see what is there
and what you might need to add or change, prepare the data to get it ready for analysis, analyze and
report on the data, and export results to various report and data formats.

SQL Procedure

SAS tables

SAS tables
PROC
Views
INPUT OUTPUT
Views
SQL
DBMS tables

DBMS tables

28 Reports
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To read unstructured data, use the SAS DATA step.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-15

1.03 Multiple Choice Question

Wha t i s your experi ence wi th SQL?
a . ma i ntaining programs
b. wri ting progra ms
c. none a t a ll

29
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Procedure

PROC SQL <options>; The SQL procedure is initiated

with a PROC SQL statement.
SELECT col-name, col-name
FROM input-table
<WHERE clause>
<GROUP BY clause> The SQL procedure
<HAVING clause> is terminated with
<ORDER BY clause>; a QUIT statement.

QUIT;

30
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Multiple statements can be included in a PROC SQL step. Each statement defines a process and is
executed immediately.

SELECT Statement in PROC SQL

PROC SQL <options>;

A SELECT statement or query
query
consists of clauses and ends with
SELECT col-name, col-name
a semicolon.
FROM input-table
<WHERE clause>
<GROUP BY clause> Think of a query as
<HAVING clause> asking your data a
<ORDER BY clause>; question.

QUIT;

31
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The SELECT statement is the most commonly used SQL statement and is usually referred to as a
query.

SELECT Statement Clauses in PROC SQL

The SELECT clause lists the columns to
PROC SQL <options>; appear in the results, in the order
written.
SELECT col-name, col-name
FROM input-table The FROM clause specifies the data
<WHERE clause> sources.
<GROUP BY clause>
<HAVING clause> The SELECT statement
<ORDER BY clause>; must contain SELECT
and FROM clauses.
QUIT;

32
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-17

SELECT Statement Clauses in PROC SQL

PROC SQL <options>;

SELECT col-name, col-name

FROM input-table
The remaining
<WHERE clause>
clauses are
<GROUP BY clause>
optional.
<HAVING clause>
<ORDER BY clause>;

QUIT;
If present, the other clauses
must be in this order.

33
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When you construct a SELECT statement, you must specify the clauses in the following order:
• The SELECT clause selects columns.
• The FROM clause selects one or more source tables or views .
• The WHERE clause enables you to filter rows of data.
• The GROUP BY clause enables you to process data in groups.
• The HAVING clause works with the GROUP BY clause to filter grouped results .
• The ORDER BY clause specifies the order of rows.

Additional Statements in PROC SQL

PROC SQL <options>; Visit the SAS

documentation for all
statement(s); available statements.

QUIT;

PROC SQL supports additional statements.

Each statement begins with the keyword
and ends with a semicolon.

34
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the available PROC SQL statements, see "SQL Procedure" in the SAS ®
9.4 SQL Procedure User’s Guide documentation. You can also find the direct link in the Course
Links section on the ELP.

Exploring Tables

Explore the
customer table.

customer

• column attributes
• preview first 10 rows

35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-19

Viewing Column Attributes

Explore
Data

DESCRIBE TABLE table-name;

proc sql;
describe table sq.customer; The DESCRIBE TABLE
quit; statement lists all the
columns in a table and
their properties.

36
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The DESCRIBE TABLE statement returns results comparable to PROC CONTENTS, which shows
the contents of a SAS table and more in-depth table information, and prints the directory of the SAS
library.

PROC CONTENTS DATA=data-set;

RUN;

For more information about the CONTENTS procedure, see “CONTENTS Procedure” in Base SAS ®
9.4 Procedures Guide, Seventh Edition. You can also find the direct link in the Course Links section
on the ELP.

Previewing the Table

Explore
Data
SELECT col-name, col-name
FROM input-table <(data set options)>;
proc sql;
select FirstName, LastName, State
from sq.customer (obs=10);
quit;

37
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To select all of a table’s columns in the order in which they were stored, specify an asterisk (*) in a
SELECT clause instead of column names.
The OBS= data set option specifies the last observation that SAS processes in a data set.
For more information about data set options, see “Dictionary of Data Set Options” in the SAS® 9.4
and SAS ® Viya® 3.4 Programming documentation. You can also find the direct link in the Course
Links section on the ELP.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-21

Exploring the customer Table

Scenario
Use PROC SQL to explore the customer table.

Files
• s101d01.sas
• customer – a SAS table that contains one row per customer.

Syntax

PROC SQL;
DESCRIBE TABLE table-name;
QUIT;

PROC SQL;
SELECT col-name, col-name
FROM input-table(OBS=n);
QUIT;

Notes
• The DESCRIBE TABLE displays each column attribute in the log.
• The SELECT statement is the primary tool of PROC SQL. You use it to identify, retrieve, and
manipulate columns of data from a table. You can also use several optional clauses within the
SELECT statement to place restrictions on a query.
• The SELECT clause lists the columns that will appear in the results. The asterisk (*) selects all
columns.
• The FROM clause specifies source tables or views.

Demo
1. Open s101d01.sas from the demos folder and find the Demo section of the program. In the first
SQL procedure step, add a DESCRIBE TABLE statement to see column attributes of the
sq.customer table. Highlight the step and run the selected code. Examine the log and results.
proc sql;
describe table sq.customer;
quit;
2. In the second SQL procedure step, add a SELECT statement and select all the columns from the
sq.customer table using an asterisk. Add the OBS=10 data set option to the table to limit the
report to 10 rows.
Note: In this example, the customer table contains more than 100,000 rows. Running the
query without the OBS= option displays all the rows and might cause system issues.
proc sql;
select *
from sq.customer(obs=10);
quit;

3. Modify the SELECT statement to select the columns FirstName, LastName, and DOB. Highlight
the step and run the selected code. Examine the log and results.
Note: PROC SQL displays the permanently assigned labels if they exist.
proc sql;
select FirstName, LastName, DOB
from sq.customer(obs=10);
quit;
4. Modify the SELECT statement to select the columns CustomerID, LastName, UserID, and
DOB. Highlight the step and run the selected code. Examine the log and results.
proc sql;
select CustomerID, LastName, UserID, DOB
from sq.customer(obs=10);
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-23

SQL Options: Controlling Processing

PROC SQL INOBS=n; PROC SQL OUTOBS=n;

Limits rows from each source table Restricts the number of rows
that contribute to a query that a query outputs

proc sql inobs=10;

select CustomerID, LastName, UserID, DOB
from sq.customer;
quit;

39
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Options: Controlling Display

controls whether the row number is displayed

PROC SQL NUMBER; as the first column in the query results

proc sql inobs=10 number;

select CustomerID, LastName, UserID, DOB
from sq.customer;
quit;

40
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about PROC SQL options, see “PROC SQL Options” in the SAS® 9.4 SQL
Procedure User’s Guide, Fourth Edition documentation. You can also find the direct link in the
Course Links section on the ELP.

1.04 Activity
Open s101a04.sas from the activities fol der a nd perform the following tasks
to expl ore the sq.customer ta bl e:
1. Remove the a s terisk a nd select only the FirstName, LastName, a nd State
col umns . Run the query. Vi ew the log a nd results .
2. Remove the OBS=10 da ta s et option a nd a dd the INOBS=10 PROC SQL
option after the PROC SQL keywords a nd before the s emi colon. Run the
query. Are the res ul ts the sa me using the INOBS=10 option? What a bout
the l og?
3. After the INOBS= opti on, a dd the NUMBER opti on. Run the query. Whi ch
col umn wa s a dded to the res ults?

41
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Comparing SQL and the DATA Step

DATA Step PROC SQL

▪ Reads unstructured data ▪ Code can be more streamlined

▪ More control of how SAS processes data ▪ Describe what you want to do, not
how to do it
Can create multiple tables in one step
SQL optimizer
Includes looping and array processing

45
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

PROC SQL is a complement to the DATA step, not a replacement for the DATA step. Sometimes
PROC SQL is the best tool to use, but in other situations, it is better to use the DATA step.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Introduction to the SQL Procedure 1-25

Syntax Summary

PROC SQL <options>;

SELECT col-name, col-name DESCRIBE TABLE table-name;
FROM input-table; Explore a Table
QUIT;
SELECT Statement

PROC SQL INOBS=n OUTOBS=n NUMBER;

input-table (OBS=n) PROC SQL Options
Data Set Options

46
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Beyond SQL Essentials

What if you want to ...

. . . l ea rn a bout SAS . . . l ea rn a bout . . . l ea rn a bout rea ding

progra mmi ng techni ques to explore uns tructured data using
es s entials? your da ta ? the DATA Step?
• Take the free e-learning • Read the SAS paper Know • Read the SAS paper Reading
course SAS Programming 1: Thy Data: Techniques for Delimited Text Files into
Essentials. Data Exploration. SAS®9.
• Read the SAS paper From
Unfriendly Text File to SAS
Dataset – Reading Character
Strings.
• Visit the SAS blog Turning text
files into SAS data sets.

47
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

1.4 Solutions
Solutions to Activities and Questions

1.01 Activity – Correct Answer

Confirm that 27
SAS tables are
created.

12
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

1.02 Activity – Correct Answer

Main folder location

%let path=s:/workshop;

libname sq "s:/workshop/data";

Updates the libname.sas program to store

the LIBNAME statement for the sq library
sq

17
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
1.04 Activity – Correct Answer
1. Remove the a s terisk a nd select only the FirstName, LastName, a nd State
col umns . Run the query. Vi ew the log a nd results .

proc sql;
select FirstName, LastName, State
from sq.customer(obs=10);
quit;
Using SAS data set
options is a SAS
enhancement.

42
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
1.04 Activity – Correct Answer
2. Are the res ults the s ame using the INOBS=10 option? Yes, the results
are the same. Wha t a bout the l og? The INOBS= option restricts the
number of rows that PROC SQL retrieves from any single source. The
results are the same, but the log returns a warning.

proc sql inobs=10;

select FirstName, LastName, State
from sq.customer;
quit;

43
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

1.04 Activity – Correct Answer

3. Whi ch col umn wa s added to the res ults? The ROW column. The
NUMBER option specifies whether the SELECT statement includes the
row number of the data as the rows are retrieved.

proc sql inobs=10 number;

select FirstName, LastName, State Options in the
from sq.customer; PROC SQL
quit; statement are SAS
enhancements.

44
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
Lesson 2 PROC SQL
Fundamentals
2.1 Generating Simple Reports .......................................................................................... 2-3
Demonstration: Creating Simple Reports.................................................................. 2-17
Demonstration: Assigning Values Conditionally ......................................................... 2-24
Practice............................................................................................................... 2-27

2.2 Summarizing and Grouping Data ............................................................................... 2-31

Demonstration: Summary Functions ........................................................................ 2-35
Demonstration: Analyzing Groups of Data ................................................................ 2-41
Demonstration: Summarizing Data Using a Boolean .................................................. 2-47
Practice............................................................................................................... 2-50

2.3 Creating and Managing Tables................................................................................... 2-53

2.4 Using DICTIONARY Tables......................................................................................... 2-63

Demonstration: Using DICTIONARY Tables .............................................................. 2-66
Practice............................................................................................................... 2-69

2.5 Solutions ................................................................................................................... 2-71

Solutions to Practices ............................................................................................ 2-71
Solutions to Activities and Questions........................................................................ 2-75
2-2 Lesson 2 PROC SQL Fundamentals

2.1 Generating Simple Reports

Creating Simple Reports

• Top 10 customers by income

• No BankID
• CreditScore greater than 700
..

Customer • Born prior to December 31, 1940

• Employed

3
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Filtering Rows Using the WHERE Clause

PROC SQL <options>;

SELECT col-name, col-name
FROM input-table
WHERE expression;
QUIT; proc sql;
select CustomerID, City, State
from sq.customer
where State='NC';
quit;

4
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The WHERE clause enables you to retrieve only the rows that satisfy a condition.
WHERE clauses can contain any of the columns in a table, including columns that are not specified
in the SELECT clause.
The WHERE clause must come after the SELECT clause and the FROM clause.

Character and Numeric Values

Standard
WHERE expression;
digits 0 – 9
minus sign
decimal point
scientific notation

Character Numeric

where State = 'NC' where Income < 30000

where State = "NC" where month(DOB) = 9

5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SAS columns can be either character or numeric.

• Character values are case sensitive and must be enclosed in single or double quotation marks.
Double quotation marks are a SAS enhancement for character strings that are typically not
allowed in DBMSs.
• Numeric values are not enclosed in quotation marks and must be standard numeric values. You
cannot include special symbols such as commas or dollar signs.

WHERE Comparison Operators

Mnemonic Symbol
LT <
GT >
EQ =
LE <=
GE >=
NE <>
¬= (EBCDIC)
^= (ASCII)

6
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Date, Time, and Datetime Values

SAS date number of days
where DOB < "01JAN2000"d since
January 1, 1960
14610
SAS time number of
where Time >= "14:45:32"t seconds since
midnight
53132
SAS datetime number of
where DateTime <= "01JAN2000 12:00:00"dt seconds since
midnight
1262347200 January 1, 1960

7
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A SAS date value is a date written in the following form: 'ddmmm<yy>yy'd or ''ddmmm<yy>yy''d.
• date='1JAN2013'd;
• date=‘01jan09’D;
A SAS time constant is a time written in the following form: 'hh:mm<:ss.s>'t or ''hh:mm<:ss.s>''t.
• time='9:25’t;
• time='9:25:19pm'T;
A SAS datetime constant is datetime value written in the following form:
'ddmmm<yy>yy:hh:mm<:ss.s>'dt or ''ddmmm<yy>yy:hh:mm<:ss.s>''dt.
• begin='01may12:9:30:00'dt;
• end='31dec13:5:00:00’dt;
• dtime='18jan2003:9:27:05am'DT;
Common Date and Time Functions:
• MONTH(date) - returns the month as a numeric value from a SAS date value.
• DAY(date) - returns the numeric day of the month from a SAS date value.
• YEAR(date) - returns the numeric year from a SAS date value.
• YRDIF(start-date, end-date, < basis>) - returns the difference in years between two dates
according to specified day count conventions.
• TODAY() - returns the current date as a numeric SAS date value.
• TIME() - returns the current time of day as a numeric SAS time value.
• DATETIME() - returns the current date and time of day as a SAS datetime value.
• DATEPART(datetime) - extracts the date from a SAS datetime value.
• TIMEPART(datetime) - extracts a time value from a SAS datetime value.
• MDY(month, day, year) - returns a SAS date value from month, day, and year values.

For more information about the SAS functions, see “SAS Functions and CALL Routines by
Category” in the SAS® 9.4 Functions and CALL Routines: Reference, Fifth Edition documentation.
You can also find the direct link in the Course Links section on the ELP.

Combining Expressions
WHERE expression-1 OR | AND expression-n;

OR AND
proc sql; proc sql;
select CustomerID, DOB select CustomerID, DOB
from sq.customer from sq.customer
where State = 'NY' or where Income > 30000 and
State = 'NC' or State = 'NC';
State = 'CA'; quit;
quit;

If any expression is true, If both expressions are

select the row. true, select the row.
8
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

IN Operator

WHERE col-name IN (value-1,…,value-n);

WHERE col-name NOT IN (value-1,…,value-n);

where State in ("NC","GA","NY")

State="NC" or State="GA" or State="NY";

The NOT
where State not in ('NC' 'GA' 'NY') operator forms a
negative
condition.

9
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The IN operator tests for values that match one of a list of values .

You can use the IN operator with character strings or numeric values to determine whether a
variable's value is among a list of values. Character values must be enclosed in quot ation marks.
Numeric values must be in standard form.

2.01 Activity
Open s102a01.sas from the activities fol der a nd perform the following tasks
to fi nd a l l customers i n the states VT, SC, or GA:
1. Compl ete the WHERE cl a use to fi lter for cus tomers i n the s tate of VT a nd
run the query.
2. Add a nother expression using the OR operator to s elect only cus tomers
from the s ta te of VT or SC. How ma ny cus tomers a re from ei ther VT or
SC?
3. Swi tch your current expres sion to use the IN operator. Add the s tate of
GA. How ma ny cus tomers are from either VT, SC, or GA?

10
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Special WHERE Operators: Missing Values

WHERE col-name IS NULL; The IS NULL operator

WHERE col-name IS NOT NULL; works for
both character and
numeric missing
values.

Numeric Character

where Income = . where Married = " "

12
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The IS NULL operator

• selects observations in which a variable has a missing value. It can be used for both character and
numeric variables.

• is ANSI standard and can be used in DBMS environments that distinguish between missing and
null values.
• is the same as, and interchangeable with, the IS MISSING operator in SAS. However, IS
MISSING is not ANSI standard.
• can be prefixed with the NOT operator to form a negative condition.
In SAS,
• a blank represents a missing character value
• a . represents a missing numeric value.

2.02 Activity
Open s102a02.sas from the activities fol der a nd perform the following
tas ks to fi nd all cus tomers with a nonmissing CreditScore va l ue tha t is
l es s tha n 500.
1. Exa mi ne the query. Add a WHERE cl a us e to fi nd a ll customers wi th a
CreditScore va l ue that i s l ess tha n 500 a nd run the query. Wha t do you
notice a bout the va l ues i n the CreditScore col umn? How ma ny rows a re
i n your report?
2. Incl ude the AND operator i n the WHERE cl a use to fi nd a ll rows that a re
l es s tha n 500 a nd not nul l. Us e a method of your choi ce. How ma ny rows
a re i n your fi na l report?

13
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Special WHERE Operators: Range of Values

WHERE col-name BETWEEN value-1 AND value-2;

The BETWEEN-AND
WHERE col-name NOT BETWEEN value-1 AND value-2; operator can be
used on character
values.

where CreditScore between 700 and 799

where CreditScore >= 700 and CreditScore <= 799

Selects rows with values between and

including the endpoints that you specify
16
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Special WHERE Operators: Pattern Matching

WHERE col-name LIKE "value";

where FirstName like "Z%"; where FirstName like "Z_l%";

Zachary Zelda
wildcard wildcard
Zelda Zulma
for any for a
Zulma number of Zelma single
Zula characters Zola character
Zoe
Zandra
17
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The LIKE operator tests for values that match a specified pattern.
• a percent sign (%) to match any number of characters
• an underscore (_) to match one arbitrary character

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-10 Lesson 2 PROC SQL Fundamentals

Sorting the Output with the ORDER BY Clause

PROC SQL <options>;

SELECT col-name, col-name
FROM input-table
WHERE expression
ORDER BY col-name <DESC> <, col-name>; The ORDER BY
QUIT; clause defines the
order in which rows
Sort Methods are displayed in the
• name or alias results.
• multiple columns
• calculated column
• integer position

18
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Unless an ORDER BY clause is included in the SELECT statement, a particular order to the output
rows (such as the order in which the rows are encountered in the queried table) cannot be
guaranteed. Without an ORDER BY clause, the order of the output rows is determined by the
internal processing of PROC SQL, the default collating sequence of SAS, and your operating
environment. Therefore, if you want your query results to appear in a particular order, use the
ORDER BY clause.
• The PROC SQL default sort order is ascending.
• When you use an ORDER BY clause, you change the order of the result s but not the order of the
rows that are stored in the source table.
• PROC SQL sorts missing values before nonmissing values. Therefore, when you specify
ascending order, missing values appear first in the query results.
• If multiple ORDER BY columns are specified, the first one determines the major sort order.

2.03 Activity
Open s102a03.sas from the activities fol der a nd perform the following tasks
to s ort the report by CreditScore a nd LastName:
1. Compl ete the ORDER BY cl a use and sort by CreditScore. Run the query
a nd exa mine the report. Wha t i s the default s ort order?
2. Add the keyword DESC a fter the CreditScore col umn i n the ORDER BY
cl a us e. Run the query a nd examine the report. Wha t does the DESC
option do?
3. Add a s econdary s ort col umn to s ort by LastName. Run the query. Who is
the fi rs t cus tomer on the report?
4. Remove LastName from the SELECT cl a us e a nd rerun the query. Are the
res ul ts s till s orted by LastName wi thi n CreditScore?

19
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Order Columns by Position

1 2 3
proc sql;
select FirstName, LastName, CreditScore
from sq.customer
where CreditScore > 830
order by 3 desc, 2;
quit;
You can also sort
by column
position.

21
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-12 Lesson 2 PROC SQL Fundamentals

Enhancing Reports

What does this

report show?

22
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Enhancing Reports
Add labels
Add title

Add
Add footnote formats
23
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Adding Titles and Footnotes

TITLE<n> "title-text"; FOOTNOTE<n> "footnote-text";

title1 "Customers from Hawaii";

footnote1 "March 6";
Adds a title and a
sql query footnote
title;
footnote;
Clears the title
and footnote

24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

TITLE is a global statement that establishes a permanent title for all reports created in your SAS
session. The syntax is the keyword TITLE followed by the title text enclosed in quotation marks. You
can have up to 10 titles. Specify a number 1 through 10 after the keyword TITLE to indicate the line
number. TITLE and TITLE1 are equivalent. FOOTNOTE follows the same rules as TITLE.

Adding Column Labels

SELECT col-name <LABEL='LABEL'> < FORMAT=formatw.d>

select FirstName, LastName, State,

UserID "Email Address",
Income label="Estimated Income" format=dollar16.2,
DOB format=date9.

25
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-14 Lesson 2 PROC SQL Fundamentals

Adding Column Formats

SELECT col-name <LABEL='LABEL'> <FORMAT=formatw.d>

select FirstName, LastName, State,

UserID "Email Address",
Income label="Estimated Income" format=dollar16.2,
DOB format=date9.

SAS displays an asterisk if the

specified width is not wide enough.

26
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Format names always contain a period (.) as a required delimiter. The w specifies the total width of
the value, including decimal places and special characters. If you do not specify a format width that
is large enough to accommodate a value, SAS automatically adjusts to display as much of the stored
value as possible.
Depending on the particular format, the d specifies the number of decimal places in the value. The d
is optional for numeric formats. For example, to display the value 1234 as $1,234 in a report, you
can use the DOLLAR6.0 or DOLLAR6. format.

2.04 Activity
Open s102a04.sas from the activities fol der a nd perform the following tasks
to enha nce a report:
1. Exa mi ne the query. Add the title "Customers from Hawaii" a nd a
footnote us ing today's date. Run the progra m and exa mine the new title
a nd footnote i n your report.
2. Appl y LABEL="Email Address" to the UserID col umn a nd LABEL=
"Estimated Income" to the Income col umn.
3. Appl y FORMAT=DATE9. to the DOB col umn a nd FORMAT=DOLLAR16.2
to the Income col umn. Run the program a nd examine the report.
4. Cha nge the DOLLAR16.2 forma t to DOLLAR7.2. Run the progra m. Wha t
ha ppens to the va l ues in the Income col umn?

27
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SAS Formats

You can use the

FORMAT procedure
to create your own
SAS Formats PROC FORMAT format.

30
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about SAS formats, see “Formats by Category” in the SAS® 9.4 Formats and
Informats: Reference documentation. You can also find the direct link in the Course Links section on
the ELP.
For more information about PROC FORMAT, see “FORMAT Procedure” in the Base SAS ® 9.4
Procedures Guide, Seventh Edition documentation. You can also find the direct link in the Course
Links section on the ELP.

Common Numeric Formats:

Format Name Example Value Format Applied Formatted Value

w.d 12345.67 5. 12346

w.d 12345.67 8.1 12345.7

COMMAw.d 12345.67 COMMA8.1 12,345.7

DOLLARw.d 12345.67 DOLLAR10.2 $12,345.67

DOLLARw.d 12345.67 DOLLAR10. $12,346

YENw.d 12345.67 YEN7. ¥12,346

EUROXw.d 12345.67 EUROX10.2 €12.345,67

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-16 Lesson 2 PROC SQL Fundamentals

Common Date Formats:

Example Value Format Applied Formatted Value

21199 DATE7. 15JAN18

21199 DATE9. 15JAN2018

21199 MMDDYY10. 01/15/2018

21199 DDMMYY8. 15/01/18

21199 MONYY7. JAN2018

21199 MONNAME. January

21199 WEEKDATE. Monday, January 15, 2018

Creating Simple Reports

Scenario
Use the WHERE and ORDER BY clauses to create three reports.

Files
• s102d01.sas
• customer – a SAS table that contains one row per customer

Syntax

TITLE<n> 'title-text’;
PROC SQL OUTOBS=n;
SELECT col-name <FORMAT=formatw.d> <LABEL='LABEL'>
FROM input-table(OBS=n)
WHERE expression
ORDER BY col-name <DESC>;
QUIT;

TITLE;

Notes
• The WHERE clause filters rows based on the expression (or expressions).
• The ORDER BY clause arranges rows based on the listed columns. The default order is
ascending. Use DESC after a column name to reverse the sort sequence.
• The OBS= data set option specifies the last rows that SAS processes from a table.
• The OUTOBS= option restricts the number of rows in the results.
• TITLE is a global statement that establishes a permanent title for all reports created in your SAS
session.
• The FORMAT= column modifier specifies a SAS format for determining how character and
numeric values in a column are displayed by the query expression.
• The LABEL= column modifier specifies a column label.

Demo
1. Open the s102d01.sas program in the demos folder and find the Demo section. Move to
Report 1. Complete the query.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-18 Lesson 2 PROC SQL Fundamentals

a. Complete the WHERE clause to filter for a missing BankID value and a value of
CreditScore greater than 700.
proc sql;
select FirstName, LastName, State,
Income, UserID
from sq.customer (obs=100)
where BankID is null and CreditScore > 700;
quit;
b. Complete the ORDER BY clause to arrange rows by descending Income.
proc sql;
select FirstName, LastName, State, Income, UserID
from sq.customer (obs=100)
where BankID is null and CreditScore > 700
order by Income desc;
quit;
c. Add the column modifiers FORMAT=DOLLAR16. to the Income column and LABEL=‘Email’
to the UserID column. Remove the OBS= data set option and add the OUTOBS=10 option in
the PROC SQL statement.
proc sql outobs=10;
select FirstName, LastName, State,
Income format=dollar16.,
UserID label='Email'
from sq.customer (obs=100)
where BankID is null and CreditScore > 700
order by Income desc;
quit;

2. Move to Report 2. Complete the query.

a. Complete the WHERE clause to filter DOB prior to 31DEC1940 and where Employed
equals Y.
proc sql;
select CustomerID, State, Zip,
DOB, UserID,
HomePhone, CellPhone
from sq.customer(obs=100)
where DOB < '31DEC1940'd and Employed='Y';
quit;

b. Complete the ORDER BY clause to arrange rows by descending DOB. Run the query and
view the results.
proc sql;
select CustomerID, State, Zip,
DOB, UserID,
HomePhone, CellPhone
from sq.customer(obs=100)
where DOB < '31DEC1940'd and Employed='Y'
order by DOB desc;
quit;
c. Add the column modifiers FORMAT= to the DOB and Zip columns. Remove the OBS= data
set option and highlight and run the query. Examine the log and results.
Note: The Z format writes standard numeric data with leading 0s. Scroll in the results and show
ZIP codes with fewer than five digits.
proc sql;
select CustomerID, State, Zip format=z5.,
DOB format=date9., UserID,
HomePhone, CellPhone
from sq.customer(obs=100)
…

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-20 Lesson 2 PROC SQL Fundamentals

Creating a New Column

expression AS alias
proc sql;
select FirstName, LastName, UserID,
yrdif(dob,'01jan2019'd) as Age
from sq.customer(obs=100);
quit;
Create a report that
contains customers
who are age
70 and over.

32
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

In addition to selecting columns that are stored in a table, you can create new columns that exist for
the duration of the query. These columns can contain text or calculations. PROC SQL writes the
columns that you create as if they were columns from the table.
The YRDIF function enables us to calculate the age of all of our customers. We begin with the start
date, which is their date of birth value. Then we use the constant of January 1, 2019, as the end
date. If you want today's date, you can use the TODAY function(). The third argument, basis,
describes how SAS calculates the date difference. We specify Age to indicate that we want the
person's age computed. We will name the column Age and run the query to create a new column
with each customer’s age.
For more information about the YRDIF function, see “YRDIF Function” in the SAS® 9.4 Functions
and CALL Routines: Reference, Fifth Edition documentation. You can also find the direct link in the
Course Links section on the ELP.

2.05 Activity
Open s102a05.sas from the activities fol der a nd perform the following tasks
to fi nd a l l customers 70 yea rs ol d a nd ol der:
1. Exa mi ne a nd run the query. Vi ew the res ults.
2. Add the expres sion yrdif(dob,'01jan2019'd) i n the SELECT cl a use a fter
UserID to crea te a new col umn. Run the query a nd examine the results.
Wha t i s the na me of the new col umn?
3. Add as Age a fter your functi on. Run the query a nd examine the results .
Wha t cha nges?
4. Remove the OBS= da ta s et option i n the FROM cl a use and add a WHERE
cl a us e to return rows where Age i s greater tha n or equal to 70. Run the
query. Di d the query run s uccessfully?

33
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subsetting Calculated Values

proc sql;
select FirstName, LastName, UserID,
yrdif(dob,'01jan2019'd) as Age
from sq.customer
where yrdif(dob,'01jan2019'd)>=70;
quit;
proc sql;
select FirstName, LastName, UserID,
yrdif(dob,'01jan2019'd) as Age
from sq.customer
where calculated Age>=70;
quit;

35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When you use a column alias to refer to a calculated value, you must use the CALCULATED
keyword with the alias to inform PROC SQL that the value is calculated within the query. You can
use an alias to refer to a calculated column in a SELECT clause, a WHERE clause, or an ORDER
BY clause. The ORDER BY clause does not require the calculated keyword.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-22 Lesson 2 PROC SQL Fundamentals

Assigning Values to a New Column Conditionally

CODE VALUE
M Married Customer Partial
S Single
D Divorced
W Widowed

CATEGORY RANGE
Excellent 750+
Good 700 – 749
Fair 650 – 699
Poor 550 – 649
Bad 550 & below
36
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Simple CASE Expression

37
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Simple CASE Expression

proc sql;
select FirstName, LastName, State, CreditScore,
case
when CreditScore >= 750 then "Excellent"
when CreditScore >= 700 then "Good"
when CreditScore >= 650 then "Fair"
when CreditScore >= 550 then "Poor"
when CreditScore >= 0 then "Bad"
else "Unknown"
end as Category
from sq.customer(obs=1000); The first WHEN clause evaluated
quit; as true determines which value
ELSE provides alternate action if the CASE expression returns.
no WHEN expressions are true.
38
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

CASE-Operand Form
proc sql;
select FirstName, LastName, State, CreditScore,
case Married
when "M" then "Married" equivalent of
when "D" then "Divorced" Married="D"
when "S" then "Single"
when "W" then "Widowed"
else "Unknown"
end as Category
from sq.customer(obs=1000); A test of equality
quit; is implied.
GT LT <= >= NE =

39
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-24 Lesson 2 PROC SQL Fundamentals

Assigning Values Conditionally

Scenario
Use the CASE expression to create columns conditionally.

Files
• s102d02.sas
• customer – a SAS table that contains one row per customer

Syntax

PROC SQL;
SELECT col-name, <col-name>,
CASE <case-operand>
WHEN condition THEN result-expression
<WHEN condition THEN result-expression>
<ELSE result-expression>
END <AS column>
FROM input-table;
QUIT;

Notes
• CASE expressions enable you to interpret and change some or all the data values in a column to
make the data more useful or meaningful.
• You can use a CASE expression anywhere that you can use a column name.

Demo
1. Open the s102d02.sas program in the demos folder and find the Demo section.
2. In the Simple Case Expression section:
a. Highlight and run the query. Examine the log and results.
b. Complete the WHEN and ELSE expressions in the simple CASE expression. Highlight and
run the query. Examine the log and results.
proc sql;
select FirstName, LastName, State, CreditScore,
case
when CreditScore >=750 then 'Excellent'
when CreditScore >=700 then 'Good'
when CreditScore >=650 then 'Fair'
when CreditScore >=550 then 'Poor'
when CreditScore >=0 then 'Bad'
end as CreditCategory
from sq.customer(obs=1000);
quit;

c. Add the ELSE expression to change remaining values to Unk nown. Highlight and run the
query. Examine the log and results.
…
else 'Unknown'
end as CreditCategory
…
d. Add a WHERE clause to filter the table for customers with Excellent credit and remove the
OBS=1000 data set option. Highlight and run the query. Examine the log and results.
proc sql;
select FirstName, LastName, State, CreditScore,
case
when CreditScore >=750 then 'Excellent'
when CreditScore >=700 then 'Good'
when CreditScore >=650 then 'Fair'
when CreditScore >=550 then 'Poor'
when CreditScore >=0 then 'Bad'
else 'Unknown'
end as CreditCategory
from sq.customer(obs=1000)
where calculated CreditCategory = 'Excellent';
quit;
3. Move to the CASE-OPERAND FORM section.
a. Highlight and run the query. Examine the log and results.
b. Complete the WHEN and ELSE expressions in the simple CASE-Operand expression.
Highlight and run the query. Examine the log and results.
proc sql;
select FirstName, LastName, State, CreditScore, Married,
case Married
when 'M' then 'Married'
when 'S' then 'Single'
when 'D' then 'Divorced'
when 'W' then 'Widowed'
else 'Unknown'
end as MarriedCategory
from sq.customer(obs=1000);
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-26 Lesson 2 PROC SQL Fundamentals

Syntax Summary
SELECT col-name, col-name TITLE<n> "title-text";
FROM input-table FOOTNOTE<n> "footnote-text";
WHERE expression FORMAT=formatw.d
ORDER BY col-name <DESC>; LABEL = 'LABEL'
Query Enhance Reports

column AS alias
CASE EXPRESSION
"01JAN2000 12:00:00"dt
WHERE col-name IS NULL Create Columns "01JAN2019"d
WHERE col-name LIKE
"14:45:32"t
WHERE col-name BETWEEN-AND
WHERE CALCULATED column SAS Dates and Times
Filter Data
41
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Practice

If you restarted your SAS session, open and submit the libname.sas program in the course files.

Level 1
1. Querying a Table
The sq.transactionfull table contains customer and transaction information. Write a PROC SQL
step to generate a report for large transactions that are not related to tuition payments to
universities. Use the following requirements as you generate the report:
a. Write a query to display the following columns in this order: CustomerName,
MerchantName, Type, Service, and Amount from the sq.transactionfull table.
1) Select rows that have a transaction Amount value greater than $1,000 and a Service
value not equal to University.
2) Order the rows such that the largest transaction is listed first.
3) Format the Amount column with the DOLLAR10.2 format.
4) Label CustomerName as Customer Name, and Amount as Transaction Amount.
5) Add the following title: Large Non-Educational Transactions
6) Run the program and review the results.
Partial Results

b. How much did Pamela Maiden spend on Home Insurance?

2. Using a Function in a Filter
The sq.customer table contains customer demographic and banking information. Write a query
to create a report for customers with May birthdays in the state of North Carolina. Use the
following requirements as you generate the report:
a. Using the sq.customer table as input, display the following columns in this order:
FirstName, LastName, DOB
1) Add a filter to subset the rows based on the month value 5 for DOB and a State value of
NC.
2) Order the rows by descending DOB.
3) Format the DOB column with DATE9.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-28 Lesson 2 PROC SQL Fundamentals

4) Add the following title: Customers with May Birthdays in North Carolina
5) Run the program and review the results.
Partial Results

b. What is Amy Murray’s date of birth?

Level 2
3. Working with Datetime Values
The sq.transactionfull table contains a list of customer and transaction information. Write a
query to create a report displaying transactions that took place in November and December of
any year. Use the following requirements as you generate the report:
a. Use the sq.transactionfull table to select the following columns in this order:
CustomerName, MerchantName, Amount.
1) Create a new column named TransactionDate by using the DATEPART function to
extract the SAS date value from the DateTime column. Format the new column using the
DATE9. format.
2) Filter the data to select rows where the month of the transaction date is November or
December and the Service value is not equal to University.
3) Order the report by the original DateTime column.
4) Format the Amount column with DOLLAR10.2.
5) Label CustomerName as Customer Name, MerchantName as Merchant Name,
Amount as Transaction Amount, and TransactionDate as Transaction Date.
6) Add the following title: November/December Transactions
7) Run the program and review the results.

Partial Results

b. What value of Merchant Name is on the first documented transaction in December?

Challenge
4. Conditional Processing with a Dynamic Title
Write a query to conditionally create a new column based on a customer’s age. Use the following
requirements as you generate the report:
a. Using the sq.customer file as input, use a character format to display the first initial of
FirstName labeled as Initial and to display LastName labeled as Last Name. In addition,
display CreditScore with the label Credit Score.
1) Create a new column named Generation based on the customer’s date of birth (DOB),
using the following logic:
a) DOB between 01JAN1928 and 31DEC1945 Generation = Silent.
b) DOB between 01JAN1946 and 31DEC1964 Generation = Boomer.
c) DOB between 01JAN1965 and 31DEC1979 Generation = GenX.
d) DOB between 01JAN1980 and 31DEC1996 Generation = Millennial.
e) DOB on or after 01JAN1997 Generation = Post-Millennial.
f) Otherwise, set Generation = Unk nown.
2) Filter the data to select rows where CreditScore is not missing and State = VT.
3) Order the report by descending CreditScore within each value of Generation.
4) Add a dynamic title that specifies the exact date the report was run: Created on
Dynamic Date Value (Hint: Research the %QSYSFUNC macro function). The date
value in the title shown below is formatted using the WEEKDATE format.
5) Run the program and review the results.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-30 Lesson 2 PROC SQL Fundamentals

Partial Results (The date in the title changes based on when the program is run.)

b. Is the date in your title today's date?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-31

2.2 Summarizing and Grouping Data

Eliminating Duplicate Rows

customer

What unique State

values occur in the
customer table?

44
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Distinct Keyword

PROC SQL;
SELECT DISTINCT col-name <,col-name> The DISTINCT keyword applies to
FROM input-table all columns in the SELECT list.
QUIT;
proc sql;
select distinct State
from sq.customer;
quit;

45
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Use the DISTINCT keyword to eliminate duplicate rows. The DISTINCT keyword applies to all
columns in the SELECT list. One row is displayed for each unique combination of values.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-32 Lesson 2 PROC SQL Fundamentals

When you specify all of a table's columns in a SELECT clause with the DISTINCT keyword, PROC
SQL eliminates duplicate rows, or rows in which the values in all of the columns match, from the
results.

2.06 Activity
Open s102a06.sas from the activities fol der a nd perform the following tasks
to el i minate duplica te va lues i n a table:
1. Exa mi ne a nd run the query. Vi ew the res ults.
2. Cha nge the State col umn i n the SELECT cl a use to the Employed col umn.
Run the query. Wha t does this query s how?
3. Add the Married col umn i n the SELECT cl a use a fter the Employed
col umn. Run the query. Wha t does this query s how?

46
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Summarizing Data
statepopulation

summarize data

48
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-33

Summary Functions: Down a Column

SELECT summary function(column);

proc sql;
select max(PopEstimate1) as MaxEst format=comma16.,
min(PopEstimate1) as MinEst format=comma16.,
avg(PopEstimate1) as AvgEst format=comma16.
from sq.statepopulation;
quit;

49
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When you use a summary function with a single argument, nonmissing values are totaled down a
column.

Summary Functions: Across a Row

SELECT summary function(column1, column-n);

proc sql;
select Name, PopEstimate1, PopEstimate2, PopEstimate3,
max(PopEstimate1, PopEstimate2, PopEstimate3)
as MaxEst format=comma16.
from sq.statepopulation;
quit;

50
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A summary function with multiple arguments, nonmissing values are totaled across a row.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-34 Lesson 2 PROC SQL Fundamentals

Commonly Used Summary Functions

SQL SAS Returned Value

AVG MEAN Mean (average) value

COUNT FREQ, N Number of nonmissing values

MAX MAX Largest value

MIN MIN Smallest nonmissing value

SUM SUM Sum of nonmissing values

NMISS Number of missing values

STD Standard deviation

VAR Variance

51
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When more than one argument is used within an SQL aggregate function, the function is no longer
considered to be an SQL aggregate or summary function. If there is a like-named Base SAS
function, then PROC SQL executes the Base SAS function, and the results that are returned are
based on the values for the current row. If no like-named Base SAS function exists, then an error
occurs. For example, if you use multiple arguments for the AVG function, an error occurs because
there is no AVG function for Base SAS.
For more information about SAS functions, see the SAS® 9.4 Functions and CALL Routines:
Reference, Fifth Edition documentation. You can also find the direct link in the Course Links section
on the ELP.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-35

Summary Functions

Scenario
Use summary functions to analyze a table.

Files
• s102d03.sas
• statepopulation – a SAS table that contains forecasted US population information by states

Syntax

SELECT summary function(column);

SELECT summary function(column1, column-n);

Notes
• When you use a summary function with multiple arguments, nonmissing values are totaled
across a row.
• When you use a summary function with a single argument, nonmissing values are totaled down
a column.

Demo
1. Open the s102d03.sas program in the demos folder and find the Demo section. Highlight and
run the DESCRIBE TABLE statement and query in the Explore the sq.statepopulation table
section. Examine the log and results.
proc sql inobs=10;
describe table sq.statepopulation;
select Region, Division, Name, PopEstimate1, PopEstimate2,
PopEstimate3
from sq.statepopulation;
quit;
2. Move to Method 1 – Down a Column. In the query complete the following:
a. Run the query and examine the results.
proc sql;
select count(PopEstimate1) as TotalStates,
mean(PopEstimate1) as Mean format=comma16.
from sq.statepopulation;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-36 Lesson 2 PROC SQL Fundamentals

b. In the SELECT clause, add three columns to find the standard deviation, minimum, and
maximum of PopEstimate1 using the STD, MIN, and MAX functions. Use the COMMA16.
format for all new columns. Highlight and run the query. Examine the log and results.
proc sql;
select count(PopEstimate1) as TotalStates,
mean(PopEstimate1) as Mean format=comma16.,
std(PopEstimate1) as StdDev format=comma16. ,
min(PopEstimate1) as Min format=comma16.,
max(PopEstimate1) as Max format=comma16.
from sq.statepopulation;
quit;

c. Move to SAS Method – PROC MEANS below the query. SAS has procedures to do similar
summarization. Highlight and run the MEANS procedure. Examine the log and results.
proc means data=sq.statepopulation maxdec=0;
var PopEstimate1;
run;

3. Move to Method 2 – Across a Column. In the query, complete the following:

a. The new column named Mean generates the average population estimate for the next three
years for each state using the AVG function. In addition, the MIN and MAX functions are
used to create new columns that generate the minimum and maximum population estimate.
Highlight and run the query. Examine the log and results.
Note: When more than one argument is used within an SQL aggregate function, the function is
no longer considered to be an SQL aggregate or summary function. This causes an
error.
proc sql;
select Name,
PopEstimate1 format=comma16.,
PopEstimate2 format=comma16.,
PopEstimate3 format=comma16.,
avg(PopEstimate1, PopEstimate2, PopEstimate3) as Mean
format=comma16.,
min(PopEstimate1, PopEstimate2, PopEstimate3) as Min
format=comma16.,
max(PopEstimate1, PopEstimate2, PopEstimate3) as Max
format=comma16.
from sq.statepopulation;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-37

b. Replace the AVG function with the MEAN function. Highlight and run the query. Examine
the log and results.
…
mean(PopEstimate1, PopEstimate2, PopEstimate3) as Mean
format=comma16.
…
c. In the MAX function, change the arguments to of PopEstimate1-PopEstimate3. Highlight
and run the query. Examine and discuss the syntax error.
…
mean(PopEstimate1, PopEstimate2, PopEstimate3) as Mean
format=comma16.,
min(PopEstimate1, PopEstimate2, PopEstimate3) as Min
format=comma16. ,
max(of PopEstimate1-PopEstimate3) as Max
format=comma16.
…

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-38 Lesson 2 PROC SQL Fundamentals

Summarizing Data Using the COUNT Function

An asterisk
specifies all rows.

SELECT COUNT(argument);
proc sql;
select count(*) as TotalCustomers format=comma12.
from sq.customer;
quit;

53
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The COUNT function counts the number of rows returned by a query.

• The * (asterisk) counts all rows in a table or group.
• A column name as the argument counts the number of nonmissing values in that column.

2.07 Activity
Open s102a07.sas from the activities fol der a nd perform the following tasks
to s umma ri ze a table using the COUNT function:
1. Exa mi ne a nd run the query. Vi ew the res ults. Why i s the va lue of
MaritalStatus di fferent from the va lue of TotalRows?
2. Ins i de the COUNT function, a dd the DISTINCT keyword in front of the
Married col umn a nd run the query. Wha t does the new report s how?

54
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-39

Grouping Data

How many What is the average

customers are in credit score of those
each state? customers?
..

customer

56
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Grouping Data
SELECT col-name, summary function(column)
The GROUP BY clause summarizes
FROM input-table
groups of data by a specified WHERE expression
column or columns. GROUP BY col-name <,col-name>
ORDER BY col-name DESC;

select State, count(*) as TotalCustomers format=comma7.

from sq.customer
where BankID is not null
group by State
order by TotalCustomers desc;

57
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You can use the GROUP BY clause to do the following:

• classify the data into groups based on the values of one or more columns .
• calculate statistics for each unique value of the grouping columns .
• group by multiple columns, separating the column names with commas within the GROUP BY
clause. You can use aggregate functions with any of the columns that you select.
When you use a GROUP BY clause without an aggregate function, PROC SQL treats the GROUP
BY clause as if it were an ORDER BY clause, displaying a corresponding mess age in the log.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-40 Lesson 2 PROC SQL Fundamentals

Filtering Grouped Data

SELECT col-name, summary function(column
The HAVING clause instructs FROM input-table
PROC SQL how to filter the data WHERE expression
after the data is summarized. GROUP BY col-name <,col-name>
HAVING expression
ORDER BY col-name DESC;

select State, count(*) as TotalCustomers format=comma7.

from sq.customer
where BankID is not null
group by State
having TotalCustomers > 6000
order by TotalCustomers desc;

58
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You can use a HAVING clause with a GROUP BY clause to filter grouped data. The HAVING clause
affects groups in a way that is similar to how a WHERE clause affects individual rows. When you
use a HAVING clause, PROC SQL displays only the groups that satisfy the HAVING expression.

Because the WHERE clause is evaluated before a row is available for processing and determines
which individual rows are available for grouping, you cannot use a WHERE clause to subset
grouped rows by referring to the calculated summary column.
You can still use the WHERE clause with the HAVING clause. The WHERE clause filters the
rows coming from the input table or tables, and the HAVING clause filters the aggregated data
post-summarization.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-41

Analyzing Groups of Data

Scenario
Use the GROUP BY clause to group data and produce summary statistics for each group.

Files
• s102d04.sas
• customer – a SAS table that contains US population information by country, region, and state

Syntax

PROC SQL;
SELECT col-name, col-name
FROM input-table
WHERE expression
GROUP BY col-name
HAVING expression
ORDER BY col-name <DESC>;
QUIT;

Notes
• The GROUP BY clause enables you to break query results into subsets of rows. When you use
the GROUP BY clause, you use an aggregate function in the SELECT clause
• The HAVING clause is a “WHERE” clause for grouped data.

Demo
1. Open the s102d04.sas program in the demos folder and find the Demo section. Notice that the
query creates a report of the State column in the customer table and limits the output to 1,000
rows. Highlight and run the query. Examine the log and results.
Note: The table is limited for development purposes because the customer table has more
than 100,000 rows. After we finalize our query, we can run it on the entire table.
Note: When you use a GROUP BY clause without an aggregate function, PROC SQL treats
the GROUP BY clause as if it were an ORDER BY clause, displaying a corresponding
message in the log.
proc sql;
select State
from sq.customer(obs=1000)
group by State;
quit;
2. Modify the query to count the total number of customers in each state by using the COUNT
function. Name the column TotalCustomers. Highlight and run the query. Examine the log and
results.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-42 Lesson 2 PROC SQL Fundamentals

proc sql;
select State, count(*) as TotalCustomers format=comma7.
from sq.customer(obs=1000)
group by state;
quit;
3. Add an ORDER BY clause to the query to sort the report by descending TotalCustomers.
Remove the OBS=1000 data set option and run the final query. Examine the log and results.
proc sql;
select State, count(*) as TotalCustomers format=comma7.
from sq.customer(obs=1000)
group by state
order by TotalCustomers desc;
quit;
4. Replace State in the SELECT and GROUP BY clauses with BankID. Highlight and run the
query. Examine the log and the results.
Note: Missing values are grouped and summarized.
proc sql;
select BankID, count(*) as TotalCustomers format=comma7.
from sq.customer
group by BankID
order by TotalCustomers desc;
quit;
5. Add the Employed column after BankID in the SELECT and GROUP BY clauses. Highlight and
run the query. Examine the log and the results.
proc sql;
select BankID, Employed,
count(*) as TotalCustomers format=comma7.
from sq.customer
group by BankID, Employed
order by TotalCustomers desc;
quit;
6. Add a WHERE clause to filter for TotalCustomers greater than 10,000. Highlight and run the
query. Examine the log and the results.
Note: Because the WHERE clause is evaluated before a row is available for processing and
determines which individual rows are available for grouping, you cannot use a WHERE
clause to subset grouped rows by referring to the calculated summary column.
proc sql;
select BankID, Employed,
count(*) as TotalCustomers format=comma7.,
avg(CreditScore) as avgCreditScore format=3.
from sq.customer
where calculated TotalCustomers > 10000
group by BankID, Employed
order by TotalCustomers desc;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-43

7. Remove the WHERE clause and insert a HAVING clause below the GROUP BY clause.
Highlight and run the query. Examine the log and the results.
Note: The order of the clauses is required.
proc sql;
select BankID, Employed,
count(*) as TotalCustomers format=comma7.
from sq.customer
where calculated TotalCustomers > 10000
group by BankID, Employed
having TotalCustomers > 10000
order by TotalCustomers desc;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-44 Lesson 2 PROC SQL Fundamentals

Extracting Data from a Datetime Value

DATEPART(datetime-value) TIMEPART(datetime-value)

select DateTime,
datepart(DateTime) as Date format=date9.,
timepart(DateTime) as Time format=time.,
Amount
from sq.transaction;
Extract the date and
time values from the
DateTime column.

60
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Summarizing Data by Month

select month(datepart(DateTime)) as Month,

Median(Amount) as MedianSpent format=dollar16.
from sq.transaction
group by Month;

Wrap the DATEPART function inside the

MONTH function to extract the numeric month.

61
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-45

2.08 Activity
Open s102a08.sas from the activities fol der a nd perform the following tasks
to s umma ri ze data us ing date functions:
1. Exa mi ne a nd run the query. Vi ew the res ults. Whi ch month has the
hi ghest va lue for MedianSpent?
2. Repl ace the MONTH function with the QTR function. Cha nge the name
of the Month col umn to Qtr. Run the query. Wha t i s the error?
3. Repl ace Month i n the GROUP BY cl a us e wi th Qtr. Run the query. Whi ch
qua rter ha s the hi ghest va lue for MedianSpent?

62
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Counting Rows That Meet a Specified Criterion

customer

Create a table that

retrieves the number of
customers in each state
who are under 25 and
over 64.
customercount

64
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-46 Lesson 2 PROC SQL Fundamentals

Using Boolean Expressions

If Age is less than 25, the

value is 1 (true).
create table CustomerCount as
select State,
yrdif(DOB,"01JAN2020"d,'age')<25 as Under25
from sq.customer;

If Age is greater than 25, the

value is 0 (false).

65
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Boolean expressions evaluate to true (1) or false (0).

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-47

Summarizing Data Using a Boolean

Scenario
Use functions to summarize the number of customers under 25 and over 64 for each state.

Files
• s102d05.sas
• customer – a SAS table that contains customer information

Syntax

YRDIF(start-date, end-date <,’basis’>)

Notes
YRDIF Function
• The start date and end date specify a SAS data value.
• The basis value identifies a character constant or variable that describe how SAS calculates a
date difference.
• If you do not specify a third argument, Age becomes the default value for basis.
• The Age basis specifies that a person’s age is computed.

Demo
1. Open the s102d05.sas program in the demos folder and find the Demo section. Run the query
and examine the results.
proc sql inobs=1000;
create table CustomerCount as
select State,
yrdif(DOB,'01JAN2020'd,'age') as Age
from sq.customer;
quit;
2. Add the less than operator after the YRDIF function test if the row is under 25 years old. Rename
the column Under25. Run the query and examine the results.
proc sql inobs=1000;
create table CustomerCount as
select State,
yrdif(DOB,'01JAN2020'd,'age')<25 as Under25
from sq.customer;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-48 Lesson 2 PROC SQL Fundamentals

3. Copy the expression. Replace the < comparison operator with the > comparison operator.
Change the value from 25 to 64 and the name from Under25 to Over64. Run the query and
examine the results.
proc sql inobs=1000;
create table CustomerCount as
select State,
yrdif(DOB,'01JAN2020'd,'age')<25 as Under25,
yrdif(DOB,'01JAN2020'd,'age')>64 as Over64
from sq.customer;
quit;
4. Summarize the data by wrapping each new column with the SUM function to add all the values
of 1 to count the number of customers. Add a GROUP BY clause with the State column.
Remove the INOBS= option. Run the query and examine the results.
proc sql inobs=1000;
create table CustomerCount as
select State,
sum(yrdif(DOB,'01JAN2020'd,'age')<25) as Under25,
sum(yrdif(DOB,'01JAN2020'd,'age')>64) as Over64
from sq.customer
group by State;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-49

Syntax Summary

SELECT DISTINCT col-name, col-name WHERE expression

Filtering Rows
Eliminate Duplicates
HAVING expression
SELECT col-name, summary function(column) Filtering Summarized Data
FROM input-table
GROUP BY col-name;
GROUP BY COUNT(*)
COUNT(col-name)
SELECT summary function(column) COUNT(DISTINCT col-name)
SELECT summary function(column1, column2) COUNT Function
Summarize Data
67
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-50 Lesson 2 PROC SQL Fundamentals

Practice

Level 1
5. Eliminating Duplicates
The sq.globalfull table contains estimated financial information by geographic region and
country for the population age 15 years and older.
a. Write a query to generate a report that displays the unique CountryCode values from
sq.globalfull.
1) Order the report by the CountryCode.
2) Add the title Unique Country Codes.
3) Run the program and review the results.
Partial Results

b. Modify the query to produce a count of the unique CountryCode values.

1) Name the result of the count CountryCount.
2) Add the title Count of Unique Country Codes.
3) Run the program and review the results.
Results

c. How many unique country codes are in the sq.globalfull table?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.2 Summarizing and Grouping Data 2-51

Level 2
6. Grouping and Summarizing Data
Create a new program.
a. Write a query to generate a report that identifies which customers have the greatest
percentage of suspiciously large transactions (over $500). Use the following requirements as
you generate the report:
1) Using the sq.transactionfull table as input, select and group the report by CustomerID.
2) Create the following columns:
a) TotalTransactions using the COUNT(*) function to count the number of transactions
for each value of CustomerID.
b) SuspiciousTransactions as SUM(Amount >= 500) to count the number of
transactions greater than 500.
c) PCTSuspicious by dividing SuspiciousTransactions by TotalTransactions.
Format the new column with PERCENT8.2.
3) Select only transactions where the Service value is not equal to University.
4) Filter the output to display only summary rows where PCTSuspicious > .05.
5) Order the report by descending PCTSuspicious.
6) Add the title Customers with High Percentage of Suspicious Transactions.
7) Run the program and review the results.
Results

b. Which customer ID had the highest percentage of suspicious transactions?

Challenge
7. Grouping and Summarizing Data with Calculations
The sq.globalfull table contains estimated financial information by geographic region and
country for the population age 15 years and older.
a. Write a query to generate a report that displays how many people in each region ages 15 or
older will borrow for health or medical purposes next year and three years from now. The
report will include the one-year and three-year forecasted counts and a percent increase
between the two.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-52 Lesson 2 PROC SQL Fundamentals

1) Use the sq.globalfull table and select and group by Region.

2) Calculate three new columns.
a) The estimated number of people who will borrow for health or medical purposes by
finding the mean of EstYear1Pct, converting it to a percentage by dividing by 100,
and multiplying it by the sum of EstYear1Pop.
(1) Name the column EstCount1 and format it using commas.
(2) Use this expression as a guide: mean(EstYear1PCT)/100)*sum(E stYear1Pop)
b) The estimated number of people who will borrow for health or medical purposes by
finding the mean of EstYear3Pct, converting it to a percentage by dividing by 100,
and multiplying it by the sum of EstYear3Pop.
(1) Name the column EstCount3 and format it using commas.
(2) Hint: Copy and modify the expression used in step 1 to refer to the three-year
population estimate columns.
c) Subtract EstCount1 from EstCount3 and divide the result by EstCount1 to
determine the percent increase. Name the column PctIncrease and format it using
the PERCENT format.
3) Filter for rows where IndicatorName is Borrowed for health or medical purposes
(% age 15+).
4) Order by descending PctIncrease.
5) Add an appropriate title.
6) Run the program and review the results.
Results

b. Which region had the largest percent decrease from year 1 to year 3?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.3 Creating and Managing Tables 2-53

2.3 Creating and Managing Tables

Creating Tables

query

PROC SQL Create and

populate a table

existing table

PROC SQL Create an

empty table
table idea

70
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Tables Created from a Query

Table DATA PROC

Step Steps

71
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-54 Lesson 2 PROC SQL Fundamentals

Creating a Table from a Query Result

CREATE TABLE table-name AS query
proc sql;
create table work.highcredit as
select FirstName, LastName,
UserID, CreditScore
from sq.customer
where CreditScore > 700;
quit; work.highcredit

NOTE: Table WORK.HIGHCREDIT created, with 26006 rows and 4

columns.

72
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To create a table from a query result, use a CREATE TABLE statement with the AS keyword, and
place it before the SELECT statement. When a table is created this way, its data is derived from the
table or view that is referenced in the query's FROM clause. The new table's column names are as
specified in the query's SELECT clause list. The new table’s column attributes (the type, length,
informat, format, and extended attributes) are the same as the selected source columns.
You can create only one table in a PROC SQL step. Use the DATA step if you want to create
multiple output tables in a single step:

DATA output-table1 output-table<n>;

SET input-table;
<additional SAS statements>;
RUN;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.3 Creating and Managing Tables 2-55

2.09 Activity
Open s102a09.sas from the activities fol der a nd perform the following
tas ks :
1. Exa mi ne a nd run the query i n the Create a Table from a Query s ecti on.
Vi ew the res ults.
2. Add the CREATE TABLE s tatement a nd create a table na med Top5States.
Run the query a nd confi rm tha t the ta ble wa s created s uccessfully.
3. Run the code bel ow your SQL query. Wha t di d the code produce?

73
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copying the Structure of an Existing Table

CREATE TABLE table-name Creates a copy of

LIKE existing-table; the table
proc sql; structure
create table work.highcredit
like sq.customer(keep=FirstName LastName
UserID CreditScore);
quit;

NOTE: Table WORK.HIGHCREDIT created, with 0 rows and 4 columns.

76
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To create an empty table that has the same columns and attributes as an existing table or view, use
the LIKE clause in the CREATE TABLE statement.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-56 Lesson 2 PROC SQL Fundamentals

Creating a Table by Defining Columns

CREATE TABLE table-name
(column-name type(length)
<, ...column-name type(length)>);

proc sql; Specify column

create table work.employee names and
(FirstName char(20), attributes.
LastName char(20),
DOB date format=mmddyy10.,
EmpID num format=z6.);
quit;

NOTE: Table WORK.EMPLOYEE created, with 0 rows and 4

columns.
77
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You can create a new table without rows by using the CREATE TABLE statement to define the
columns and their attributes. You can specify a column's name, type, length, informat, format, and
label. The table definition is enclosed in parentheses. Individual column definitions are separated by
commas.

Creating Tables with PROC SQL: Summary

Description Syntax

Creating a Table from a Query Result CREATE TABLE table-name AS

SELECT…;
Copying the Structure from an Existing Table CREATE TABLE table-name
LIKE old-table-name;

Creating a Table by Defining Columns CREATE TABLE table-name

(column-name type(length)
<, ...column-name type(length)>);

78
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the CREATE TABLE statement, see “CREATE TABLE Statement” in the
SAS® 9.4 SQL Procedure User’s Guide, Fourth Edition documentation. You can also find the direct
link in the Course Links section on the ELP.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.3 Creating and Managing Tables 2-57

Inserting Rows into Tables

Use the INSERT

statement to
insert data values
into tables.

79
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

After tables are created, you can use the INSERT statement to insert data values into tables. You
can use an empty table or a table that is already populated.

Inserting Rows with a Query

INSERT INTO table-name <(column list)>
SELECT columns
FROM table-name;

proc sql;
insert into work.highcredit
(FirstName, LastName, UserID, CreditScore)
select FirstName, LastName,
UserID, CreditScore
from sq.customer
where CreditScore > 700; Columns from the query must be in
quit; the same position as in the INSERT
column list.
80
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-58 Lesson 2 PROC SQL Fundamentals

Inserting Rows with the VALUES Clause

INSERT INTO table-name <(column list)>

VALUES (value,value,...);

proc sql;
insert into employee
(FirstName, LastName, DOB, EmpID)
values("Diego", "Lopez", "01SEP1980"d, 1280)
values("Omar", "Fayed", "21MAR1989"d, 1310);
quit;

Data values align with column names

in the INSERT column list.

81
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

With the VALUES clause, you assign values to a column by position.

Note the following features of VALUES clauses:
• As with other SQL clauses, use commas to separate columns. In addition, you must use a
semicolon after the last VALUES clause only.
• If you omit data for a column without indicating a missing value, then you receive an error
message and the row is not inserted.
• To specify that a value is missing, use a space in single quotation marks for character values and
a period for numeric values.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.3 Creating and Managing Tables 2-59

Inserting Rows with the SET Clause

INSERT INTO table-name

SET column-name=value,
column-name=value,...;

proc sql;
insert into employee
set FirstName= "Diego",
LastName= "Lopez",
DOB = "01SEP1980"d, Columns within the SET clause must
EmpID = 1280; exist in the table.
quit;

82
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

With the SET clause, you assign values to columns by name. The columns can appear in any order
in the SET clause.
Note the following features of SET clauses:
• As with other SQL clauses, use commas to separate columns. In addition, you must use a
semicolon after the last SET clause only.
• If you omit data for a column, then the value in that column is a missing value.
• To specify that a value is missing, use a blank in single quotation marks for character values and a
period for numeric values.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-60 Lesson 2 PROC SQL Fundamentals

Inserting Rows with PROC SQL: Summary

Description Syntax
A query returning multiple rows based INSERT INTO table-name
on positional values <(column list)>
SELECT columns
FROM table-name;
One clause per row using positional INSERT INTO table-name
values <(column list)>
VALUES (value1,value2,...);
One clause per row using column- INSERT INTO table-name
value pairs SET column-name=value,
column-name=value,...;

83
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the INSERT Statement, see “INSERT Statement” in the SAS® 9.4 SQL
Procedure User’s Guide, Fourth Edition documentation. You can also find the direct link in the
Course Links section on the ELP.

Dropping Tables in SQL

DROP TABLE table-name;

proc sql; This is useful if you are

drop table work.employee; working with DBMSs that do
quit; not allow you to overwrite
existing tables.
NOTE: Table WORK.EMPLOYEE has
been dropped.

84
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2.3 Creating and Managing Tables 2-61

2.10 Activity
Open s102a10.sas from the activities fol der a nd perform the following tasks
to crea te a new table a nd i nsert rows into i t:
1. Exa mi ne the CREATE TABLE s tatement a nd run the query onl y. Confi rm
tha t a n empty tabl e wa s created.
2. In the Inserting Rows with a Query s ection, enter the correct col umn
na mes to complete the INSERT INTO s tatement. Run the query. How
ma ny rows were i nserted i nto the table highcredit?
3. In the Inserting Rows with the SET Clause s ection, compl ete the INSERT
INTO s tatement wi th the SET cl a use a nd insert yours elf a s a customer
i nto the highcredit ta bl e. Run the query. Wha t does the note in the l og
s a y?
4. Compl ete the code to drop the highcredit tabl e.
85
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Additional Statements

Statement Description
ALTER TABLE Adds col umns to, drops columns from, a nd changes
col umn a ttri butes i n a n existing table
UPDATE Modi fi es a col umn's va lues in existing rows of a
ta bl e or vi ew
DELETE Removes one or more rows from a ta ble or vi ew that
i s s peci fied i n the FROM cl a use

88
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the ALTER TABLE, UPDATE, or DELETE Statement, see “SQL
Procedure Reference” in the SAS® 9.4 SQL Procedure User’s Guide, Fourth Edition documentation.
You can also find the direct link in the Course Links section on the ELP.
A SAS method to alter table metadata is the DATASETS procedure. For more information about
PROC DATASETS, see “DATASETS Procedure” in the Base SAS ® 9.4 Procedures Guide, Seventh
Edition documentation. You can also find the direct link in the Course Links section on the ELP.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-62 Lesson 2 PROC SQL Fundamentals

Syntax Summary

DROP TABLE table-name; CREATE TABLE table-name AS query

Drop Tables CREATE TABLE table-name LIKE …
CREATE TABLE table-name (…)
Create Tables

INSERT INTO table-name <(column list)> query

INSERT INTO table-name <(column list)> VALUES …
INSERT INTO table-name SET …
Insert Rows

89
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.4 Using DICTIONARY Tables

Business Scenario

What tables are defined?

Which tables contain a

specific column?

Are related columns the same

type and length in all tables?

What libraries are assigned?

91
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

DICTIONARY Tables

Tables

SAS
libraries
DICTIONARY table
Columns
• information about each SAS session
• updated automatically by SAS Many
• Read-only more!
• metadata: data about other data
• valid in PROC SQL only

92
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-64 Lesson 2 PROC SQL Fundamentals

DICTIONARY tables are special Read-only PROC SQL tables or views. They retrieve information
about all the SAS libraries, SAS data sets, SAS system options, and external files that are
associated with the current SAS session. For example, the DICTIONARY.columns table contains
information such as name, type, length, and format, about all columns in all tables that are known to
the current SAS session. PROC SQL automatically assigns the DICTIONARY libref. To get
information from DICTIONARY tables, specify DICTIONARY.table-name in the FROM clause in a
SELECT statement in PROC SQL.
For more information about DICTIONARY tables, see “Accessing SAS Information by Using
DICTIONARY Tables” in the SAS® 9.4 SQL Procedure User’s Guide, Fourth Edition documentation.
You can also find the direct link in the Course Links section on the ELP.

2.11 Activity
Open s102a11.sas from the activities fol der a nd perform the following tasks
to fi nd a l l the ava ilable DICTIONARY tables in your SAS s ession:
1. Exa mi ne a nd run the program. Vi ew the l og a nd results.
2. Note the col umn l abels for the first two col umns: Member Name i s the
DICTIONARY tabl e, a nd Data Set Label i s the descri ption of tha t table.
3. Repl ace the asterisk in the SELECT cl a use a nd s elect the DISTINCT
memname a nd memlabel col umns. Run the query a nd examine a ll the
a va i lable DICTIONARY tables i n your SAS s es sion.
4. Wha t i s the da ta set label of the members DICTIONARY ta bl e?

93
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

DICTIONARY Tables

Expl ore
Da ta

DICTIONARY.tables DICTIONARY.columns DICTIONARY.libnames

96
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-66 Lesson 2 PROC SQL Fundamentals

Using DICTIONARY Tables

Scenario
Use SQL to query DICTIONARY information.

Files
• s102d06.sas
• DICTIONARY.tables, DICTIONARY.columns, DICTIONARY.libnames

Syntax

PROC SQL;
DESCRIBE TABLE DICTIONARY.<input-table>;
SELECT *
FROM DICTIONARY.<input-table>
WHERE expression
ORDER BY col-name <DESC>;
QUIT;

Notes
• Dictionary tables are Read-only metadata views that contain session metadata, such as
information about SAS libraries, data sets, and external files in use or available in the current SAS
session.
• DICTIONARY.tables provides detailed information about tables.
• DICTIONARY.columns provides detailed information about all columns in all tables.
• DICTIONARY.libnames provides detailed information about current SAS LIBNAME connections.

Demo
1. Open the s102d06.sas program in the demos folder and find the Demo section.
2. In the Explore dictionary.tables section:
a. Highlight and run the procedure. Examine the log and the results.
b. Add a WHERE clause to subset the Libname column for libraries named SQ and remove
the INOBS= option. Highlight and run the procedure. Examine the log and the results.
Note: The Libname and Memname columns are stored in all uppercase.
proc sql inobs=100;
describe table dictionary.tables;
select *
from dictionary.tables
where Libname = 'SQ';
quit;
c. Discuss the code for the SAS equivalent of DICTIONARY.tables. Highlight and run the
procedure. Examine the log and the results.

proc print data=sashelp.vtable(obs=100);

where Libname = 'SQ';
run;
3. Move to the Explore dictionary.columns section.
a. Highlight and run the procedure. Examine the log and the results.
b. Modify the WHERE clause to subset the Name column by BankID and use the UPCASE
function on the Name column. Highlight and run the procedure. Examine the log and the
results.
proc sql;
describe table dictionary.columns;
select *
from dictionary.columns
where Libname = 'SQ' and upcase(Name) = 'BANKID';
quit;
c. Discuss the code for the SAS equivalent of DICTIONARY.columns. Highlight and run the
procedure. Examine the log and the results.
proc print data=sashelp.vcolumn(obs=100);
where Libname = 'SQ';
run;
4. Move to the Explore dictionary.libnames section.
a. Highlight and run the procedure. Examine the log and the results.
b. Modify the SELECT clause by replacing the asterisks and adding the DISTINCT keyword on
the Libname column. Highlight and run the procedure. Examine the log and the results.
proc sql;
describe table dictionary.libnames;
select distinct libname
from dictionary.libnames;
quit;
c. Modify the SELECT clause by replacing distinct libname with an asterisk. Add a WHERE
clause to subset the Libname column for the SQ library. Highlight and run the procedure.
Examine the log and the results.
proc sql;
describe table dictionary.libnames;
select *
from dictionary.libnames
where Libname = 'SQ';
quit;
d. Discuss the code for the SAS equivalent of DICTIONARY.libnames. Highlight and run the
procedure. Examine the log and the results.
proc print data=sashelp.vlibnam;
where Libname = 'SQ';
run;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-68 Lesson 2 PROC SQL Fundamentals

Syntax Summary

FROM DICTIONARY.table dictionary.dictionaries

DICTIONARY dictionary.tables
Library dictionary.columns
dictionary.libnam es
DICTIONARY Tables

98
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Beyond SQL Essentials

What if you want to ...

. . . l ea rn more a bout . . . l ea rn more about

. . . l ea rn more a bout
ma na ging data us i ng DICTIONARY da ta vi s ualization
ta bl es? us i ng PROC SGPLOT?
effi ci ently?
• Read the SAS paper • View the SQL Dictionary • Visit the SAS blog series
PROC DATASETS: Tables section in the ELP for Getting started with SGPLOT.
Managing Data additional SAS papers. • Visit the SAS Data
Efficiently. Visualization blogs.
• View the SAS® 9.4
DATASETS Procedure
Tip Sheet.

99
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Practice

Level 1
8. Counting the Number of Tables in a Library
a. Write a query to create a report that displays the count of the number of tables in the SQ
library.
1) Use DICTIONARY.tables as input.
2) Name the calculated column TableCount.
3) Add an appropriate title and run the program to review the results.
Results

b. How many tables are in the sq library?

Level 2
9. Counting the Number of Tables in All Libraries
a. Write a query to create a report that displays the count of the number of tables in all libraries.
1) Use DICTIONARY.tables as input.
2) Name the calculated column TableCount.
3) Group the results by the library name.
4) Add an appropriate title and display the library name and table count as shown below.
(Your library list and counts might differ.)
Results

b. Which library has the most tables?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-70 Lesson 2 PROC SQL Fundamentals

Challenge
10. Finding Tables with a Column Using Pattern Matching
a. Write a query to create a report that displays the list of all tables in the SQ library with a
column containing ID in its name.
Display the table name and the column name containing ID as shown below.
Partial Results

b. How many unique tables have a column that contains ID in its name?

2.5 Solutions
Solutions to Practices
1. Querying a Table
/*s102s01.sas*/

title 'Large Non-Educational Transactions';

proc sql;
select CustomerName label='Customer Name',
MerchantName, Type, Service,
Amount format=dollar10.2 label='Transaction Amount'
from sq.transactionfull
where Amount >1000 and
Service ne 'University'
order by Amount desc;
quit;
title;
How much did Pamela Maiden spend on Home Insurance? $1,816.57
2. Using a Function in a Filter
/*s102s02.sas*/

title 'Customers with May Birthdays in North Carolina ';

proc sql;
select FirstName, LastName, DOB format=date9.
from sq.customer
where month(DOB)= 5 and
State='NC'
order by DOB desc;
quit;
title;
What is Amy Murray’s date of birth? 05MAY1995
3. Working with Datetime Values
/*s102s03.sas*/

title 'November/December Transactions';

proc sql;
select CustomerName label='Customer Name',
MerchantName label='Merchant Name',
Amount format=dollar10.2 label='Transaction Amount',
datepart(DateTime) as TransactionDate format=date9.
label='Transaction Date'
from sq.transactionfull
where Service ^= 'University' and
month(calculated TransactionDate) in (11, 12)
order by DateTime;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-72 Lesson 2 PROC SQL Fundamentals

quit;
title;
What value of Merchant Name is on the first documented transaction in December?
Big Burgers, Inc.
4. Conditional Processing with a Dynamic Title
/*s102s04.sas*/

title "Created on %left(%qsysfunc(today(),weekdate.))";

proc sql;
select FirstName 'First Name' format=$1.,
LastName 'Last Name',
DOB 'Date of Birth' format=date9.,
CreditScore 'Credit Score',
case
when DOB between '01jan1928'd and '31dec1945'd
then 'Silent'
when DOB between '01jan1946'd and '31dec1964'd
then 'Boomer'
when DOB between '01jan1965'd and '31dec1979'd
then 'GenX'
when DOB between '01jan1980'd and '31dec1996'd
then 'Millennial'
when DOB >= '01jan1997'd then 'Post-Millennial'
else 'Unknown'
end as Generation
from sq.customer
where CreditScore is not null and State='VT'
order by Generation, CreditScore desc;
quit;
title;
Is the date in your title today’s date? Yes
5. Eliminating Duplicates
/*s102s05.sas*/

title 'Unique Country Codes';

proc sql;
select distinct CountryCode
from sq.globalfull
order by CountryCode;
quit;

title 'Count of Unique Country Codes';

proc sql;
select count(distinct CountryCode) as CountryCount
from sq.globalfull;
quit;
title;

How many unique country codes are in the sq.globalfull table? 151
6. Grouping and Summarizing Data
/*s102s06.sas*/

title 'Customers with High Percentage of Suspicious Transactions';

proc sql;
select CustomerID,
count(*) as TotalTransactions,
sum(Amount >= 500) as SuspiciousTransactions,
calculated SuspiciousTransactions/calculated
TotalTransactions as PCTSuspicious format=percent8.2
from sq.transactionfull
where Service ^= 'University'
group by CustomerID
having PCTSuspicious>.05
order by PCTSuspicious desc;
quit;
title;
Which customer ID had the highest percentage of suspicious transactions? 1973179983
7. Grouping and Summarizing Data with Calculations
/*s102s07.sas*/

title 'Forecasted Regions who will Borrow for Health or Medical

Purpose';
proc sql;
select Region,
(mean(EstYear1PCT)/100)*sum(EstYear1Pop) as EstCount1
format=comma16.,
(mean(EstYear3PCT)/100)*sum(EstYear3Pop) as EstCount3
format=comma16.,
(calculated EstCount3 - calculated EstCount1)/calculated
EstCount1 as PctIncrease format=percent7.2
from sq.globalfull
where IndicatorName = 'Borrowed for health or medical purposes
(% age 15+)'
group by Region
order by PctIncrease desc;
quit;
title;
Which region had the largest percent decrease from year 1 to year 3? East Asia & Pacific

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-74 Lesson 2 PROC SQL Fundamentals

8. Counting the Number of Tables in a Library

/*s102s08.sas*/

title 'Count of SQ Tables';

proc sql;
select count(*) as TableCount
from dictionary.tables
where libname='SQ';
quit;
title;
How many tables are in the sq library? 27
9. Counting the Number of Tables in All Libraries
/*s102s09.sas*/

title ‘Count of All Tables’;

proc sql;
select libname, count(*) as TableCount
from dictionary.tables
group by libname;
quit;
title;
Which library has the most tables? MAPSGFK (Results might differ.)
10. Finding Tables with a Column Using Pattern Matching
/*s102s10.sas*/

title 'SQ Tables containing ID columns';

proc sql;
select distinct MemName, Name
from dictionary.columns
where Libname = 'SQ' and Name contains 'ID';
quit;
title;
How many unique tables have a column that contains ID in its name? 13

Solutions to Activities and Questions

2.01 Activity – Correct Answer

1. Compl ete the WHERE cl a use to fi lter for cus tomers i n the s tate of VT a nd
run the query.

where State = "VT"

2. How ma ny cus tomers a re from either VT or SC? 792 customers

where State = "VT" or State = "SC"

3. How ma ny cus tomers a re from either VT, SC, or GA? 2,704 customers

where State in ("VT", "SC", "GA")

11
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
2.02 Activity – Correct Answer
1. Wha t do you noti ce a bout the va lues i n the CreditScore col umn? How
ma ny rows a re i n your report? Missing values are included in the
results, for a total of 2,197 rows.

where CreditScore < 500

SAS treats missing values as

smaller than nonmissing
values.

14
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-76 Lesson 2 PROC SQL Fundamentals

2.02 Activity – Correct Answer

2. How ma ny rows a re i n your report? Missing values are not included in
the results. The new report contains 196 rows.

where CreditScore < 500 and CreditScore is not null

where CreditScore < 500 and CreditScore is not missing

where CreditScore < 500 and CreditScore ne .

equivalent statements

15
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.03 Activity – Correct Answer

1. Wha t i s the default s ort order? Ascending
2. Wha t does the DESC opti on do? Changes the sort order to descending
3. Who i s the fi rst cus tomer on the report? Donald Leyva
4. Are the res ults s till s orted by LastName wi thi n CreditScore? Yes, you
can sort by any columns in the input table even if they are not in the
SELECT clause.

select FirstName, CreditScore

from sq.customer
where CreditScore > 830
order by CreditScore desc, LastName;

20
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
2.04 Activity – Correct Answer

28
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To dynamically create a footnote or title using today’s date, use this statement:

FOOTNOTE “%LEFT(%QSYSFUNC(TODAY(),add date format))”;

2.04 Activity – Correct Answer

4. Cha nge the DOLLAR16.2 forma t to DOLLAR7.2. Wha t ha ppens to the
va l ues i n the Income col umn? The values are missing commas and
dollar signs, and they contain only one decimal point.

When the format width is not large

enough, SAS preserves the number.

29
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-78 Lesson 2 PROC SQL Fundamentals

2.05 Activity – Correct Answer

2. Wha t i s the na me of the new col umn? Without the AS keyword, the
new column does not have a name.
3. Wha t cha nges? The AS keyword specifies a column name for the new
column.
4. Di d the query run s uccessfully? No. The WHERE clause is evaluated
before the SELECT clause. Therefore, columns used in the WHERE
clause must exist in the table listed in the FROM clause.

ERROR: The following columns were not found in the

contributing tables: Age.

34
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.06 Activity – Correct Answer

2. Wha t does this query s how? It displays the unique values of N or Y in
the Employed column.

select distinct Employed

3. Wha t does this query s how? It displays unique combinations of

Employed and Married values.

select distinct Married, Employed

47
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.07 Activity – Correct Answer

1. Why i s the va l ue of MaritalStatus di fferent from the va l ue of TotalRows?
TotalRows counts all rows in the table. MaritalStatus counts every
nonmissing occurrence in the Married column.

2. Wha t does the new report s how? You can use DISTINCT with the
COUNT function to return the number of distinct, nonmissing values
from a column.
select count(*) as TotalRows format=comma10.,
count(distinct Married) as MaritalStatus format=comma10.
from sq.customer;

55
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.08 Activity – Correct Answer

1. Whi ch month ha s the highest va lue for MedianSpent? Month 2,
February
2. Wha t i s the error? The following columns were not found in the
contributing tables: Month.
select qtr(datepart(DateTime)) as Qtr,
Median(Amount) as MedianSpent format=dollar16.
from sq.transaction
Replace Month
group by Month;
with Qtr.
3. Whi ch qua rter ha s the highest va l ue for MedianSpent? Quarter 1

63
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-80 Lesson 2 PROC SQL Fundamentals

continued...
2.09 Activity – Correct Answer
1. Exa mi ne a nd run the query. Vi ew the res ults.

2. Add the CREATE TABLE s tatement a nd create a table na med Top5States.

Run the query a nd confi rm tha t the ta ble wa s created s uccessfully.

WARNING: Statement terminated early due to OUTOBS=5 option.

NOTE: Table WORK.TOP5STATES created, with 5 rows and 2 columns.

74
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.09 Activity – Correct Answer

3. Run the code bel ow your SQL query. Wha t di d the code produce?
The SGPLOT procedure created a visualization of the newly created
table.

75
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
2.10 Activity – Correct Answer
1. There are four columns and zero rows in the new highcredit table.

NOTE: Table WORK.HIGHCREDIT created, with 0 rows and 4 columns.

2. How ma ny rows were i nserted i nto the table highcredit? 26,006 rows
were inserted into work.highcredit.
insert into work.highcredit(FirstName,LastName,
UserID,CreditScore)
select FirstName, LastName,
UserID, CreditScore
from sq.customer
where CreditScore > 700;

86
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.10 Activity – Correct Answer

3. Wha t does the note i n the l og s ay?
NOTE: 1 row was inserted into WORK.HIGHCREDIT.
proc sql;
insert into highcredit
set FirstName="Panagiotis,
LastName="Styliadis",
UserID="pstyliadis",
CreditScore=800;
quit;

4.
proc sql;
drop table highcredit;
quit;

87
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
2-82 Lesson 2 PROC SQL Fundamentals

continued...
2.11 Activity – Correct Answer
1.

DICTIONARY
table names DICTIONARY
table columns

DICTIONARY
table labels

94
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

2.11 Activity – Correct Answer

3. Run the query a nd exa mine all the a vailable DICTIONARY ta bles i n your
SAS s es sion.

4. Wha t i s the da ta set label of the members DICTIONARY ta bl e?

Tables, catalogs and Views

95
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.1 Introduction to SQL Joins ............................................................................................ 3-3

3.2 Inner Joins................................................................................................................... 3-8

Demonstration: Performing an Inner Join with PROC SQL.......................................... 3-10

Demonstration: Performing an Inner Join with Four Tables .......................................... 3-16
Practice............................................................................................................... 3-23

3.3 Outer Joins ................................................................................................................ 3-28

Demonstration: Performing a Full Join with PROC SQL.............................................. 3-35

Practice............................................................................................................... 3-40

3.4 Complex Joins........................................................................................................... 3-45

Demonstration: Performing a Reflexive Join ............................................................. 3-47

3.5 Solutions ................................................................................................................... 3-54

Solutions to Practices ............................................................................................ 3-54

Solutions to Activities and Questions........................................................................ 3-60

3-2 Lesson 3 SQL Joins

3.1 Introduction to SQL Joins

Joining Tables
smallcustomer Partial smalltransaction Partial
Partial
FirstName LastName … AccountID AccountID DateTime BankID …
Gary Sienkiewicz … 1010159565 . 07MAY18:15:35:02 . …
Sergio Lefeld … 1010367330 1010159565 16SEP18:14:57:08 101010101 …
John Oliver … 2020012887 1010183063 24FEB18:17:27:42 101010101 …
Iva Bower … 3030085224 1010367330 15MAY18:17:54:21 101010101 …
1010367330 17OCT18:11:02:38 101010101 …

3
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Joining Tables
smallcustomer Partial smalltransaction Partial
FirstName LastName … AccountID AccountID DateTime BankID …
Gary Sienkiewicz … 1010159565 . 07MAY18:15:35:02 . …
Sergio Lefeld … 1010367330 1010159565 16SEP18:14:57:08 101010101 …
John Oliver … 2020012887 1010183063 24FEB18:17:27:42 101010101 …
Iva Bower … 3030085224 1010367330 15MAY18:17:54:21 101010101 …
1010367330 17OCT18:11:02:38 101010101 …

Primary key Foreign key

4
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A primary k ey is a column that contains a unique value for each row of data. A primary key cannot
contain missing values.

A foreign k ey is a column in one table that refers to the primary key in another table.

FirstName LastName … AccountID DateTime BankID …

Gary Sienkiewicz … 1010159565 16SEP18:14:57:08 101010101 …
Sergio Lefeld … 1010367330 15MAY18:17:54:21 101010101 …
Sergio Lefeld … 1010367330 17OCT18:11:02:38 101010101 …
5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Joins combine data horizontally from multiple source tables to produce either a report or an output
table. The source tables are left intact and untouched.

SQL Join Syntax

SELECT col-name, col-name

FROM table1, table2
proc sql;
select *
from sq.smallcustomer, sq.smalltransaction;
quit;
List the table names in the
8 rows 12 rows FROM clause.

6
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Default Join
smallcustomer Partial smalltransaction Partial
FirstName LastName … AccountID AccountID DateTime BankID …
Gary Sienkiewicz … 1010159565 . 07MAY18:15:35:02 . …
Sergio Lefeld … 1010367330 1010159565 16SEP18:14:57:08 101010101 …
John Oliver … 2020012887 1010183063 24FEB18:17:27:42 101010101 …
Iva Bower … 3030085224 1010367330 15MAY18:17:54:21 101010101 …
1010367330 17OCT18:11:02:38 101010101 …

By default, SQL will join every row in the smallcustomer

table with every row in the smalltransaction
8 rows x 12 rows = 96 rows

7
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When you run this query, the Cartesian product creates a report with a total of 96 rows. A Cartesian
product is rarely the result that you want when you join tables. When working with large tables, a
Cartesian product can create an unnecessarily large report or table, and even slow down system
resources.

3.01 Activity
Open s103a01.sas from the activities fol der a nd perform the following tasks
to perform a defa ult join of two tables:
1. Exa mi ne a nd run the two queries to explore the sq.smallcustomer a nd
sq.smalltransaction ta bl es. Confirm tha t the sq.smallcustomer conta i ns
8 rows a nd the sq.smalltransaction conta i ns 12 rows .
2. In the next s ecti on, l ist the sq.smallcustomer a nd sq.smalltransaction
tabl e in the FROM cl a use a nd s eparate the tables by a comma . Run the
query a nd vi ew the l og. What note do you s ee?
3. Vi ew the res ults. Na me two i ssues with the report.

8
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Types of Joins
Inner Join Returns only
rows that match

Matches and Matches and all Matches and

nonmatches nonmatches nonmatches
Left Join Full Join Right Join

11
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Joining tables is a very common requirement when working with data. There are multiple methods
available in SAS to join tables. The most common are SQL and the DATA step. This course focuses
on SQL joins. The SAS ® Programming 2 course addresses the DATA step merge.

Table Relationships
One-to-One Many-to-Many
A B C D A B C D
1 1 1 1
2 2 1 1
3 3 2 2

One-to-Many Nonmatches
A B C D A B C D
1 1 1 2
2 1 2 3
3 2 4 4

12
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

One-to-One: A single observation in one data set is related to exactly one observation in another
data set based on the values of one or more selected variables.

One-to-many: A single observation in one data set is related to more than one observation in
another data set based on the values of one or more selected variables.

Many-to-Many: Occurs when multiple records in a table are associated with multiple records in
another table.

Nonmatches: At least one observation in one data set is unrelated to any observation in another
data set based on the values of one or more selected variables.

3.2 Inner Joins

Scenario

Report showing only

customers with their
matching
transaction

14
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Inner Join Syntax

SELECT col-name, col-name

Specify the join type

in the FROM clause.

15
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .
s103d01

SQL Inner Join Syntax

SELECT col-name, col-name

FROM table1 INNER JOIN table2
ON table1.column = table2.column;
proc sql;
select FirstName, LastName, State, Income, DateTime, Amount
from sq.smallcustomer inner join sq.smalltransaction
on smallcustomer.AccountID = smalltransaction.AccountID;
quit;
Specify the join criteria in the ON clause. You can
use other comparison operators, such as the
greater than, less than, or special where operators.

16
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .
s103d01

SQL Inner Join Syntax

SELECT col-name, col-name

Qualify the column names to specify

the location of each column.
17
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .
s103d01

Performing an Inner Join with PROC SQL

Scenario

Use PROC SQL to perform an inner join between two tables.

Files
• s103d01.sas
• smallcustomer – a SAS table that contains one row per customer
• smalltransaction – a SAS table that contains one row per customer transaction

Syntax

PROC SQL;
SELECT col-name, col-name
FROM table1 INNER JOIN table2
ON table1.col-name=table2.col-name;
QUIT;

Notes
• An SQL inner join combines matching rows between two tables.
• The two tables to be joined are listed in the FROM clause separated by INNER JOIN.
• The ON expression indicates how rows should be matched.

Demo

1. Open the s103d01.sas program in the demos folder and find the Demo section of the program.
Run the queries in the Explore the Tables section to compare the columns of the
sq.smallcustomer and sq.smalltransaction tables.
proc sql;
select *
from sq.smallcustomer;
select *
from sq.smalltransaction;
quit;
2. Find the Perform the INNER JOIN section and add sq.smallcustomer and
sq.smalltransaction to the FROM clause to perform an inner join on AccountID. Qualify
AccountID columns as table-name.col-name in the ON expression only. Highlight and run the
query.
proc sql;
select FirstName, LastName, State, Income, DateTime, MerchantID,
Amount
from sq.smallcustomer inner join

sq.smalltransaction
on smallcustomer.AccountID = smalltransaction.AccountID;
quit;
3. Add the AccountID column to the query after Amount. Highlight and run the query. Examine the
log. Why does the program fail?
Note: There is an ambiguous reference. The column AccountID is in more than one table.
proc sql;
select FirstName, LastName, State, Income, DateTime, MerchantID,
Amount, AccountID
from sq.smallcustomer inner join
sq.smalltransaction
on smallcustomer.AccountID = smalltransaction.AccountID;
quit;
4. Modify the query to qualify the AccountID column in the SELECT clause. Highlight the step and
run the selected code.

Note: Because AccountID occurs in both tables, you must qualify the column with the table
name to indicate which column you want to select.
proc sql;
select FirstName, LastName, State, Income, DateTime, MerchantID,
Amount, smallcustomer.AccountID
from sq.smallcustomer inner join
sq.smalltransaction
on smallcustomer.AccountID = smalltransaction.AccountID;
quit;
5. Modify the query to include a WHERE clause to subset for customers who have a State value of
NY (New York) and an ORDER BY clause that sorts by descending Amount.
proc sql;
select FirstName, LastName, State, Income, DateTime, MerchantID,
Amount, smallcustomer.AccountID
from sq.smallcustomer inner join
sq.smalltransaction
on smallcustomer.AccountID = smalltransaction.AccountID
where State = "NY"
order by Amount desc;
quit;

Alternative SQL Inner Join Syntax

SELECT col-name, col-name

FROM table1, table2
WHERE table1.column = table2.column;
List the tables in the FROM clause.
proc sql;
select FirstName, LastName, State, Income, DateTime, Amount
from sq.smallcustomer, sq.smalltransaction
where smallcustomer.AccountID = smalltransaction.AccountID;
quit;

List the join criteria in the WHERE clause.

19
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Table Aliases

FROM table1 <AS> alias1, table2 <AS> alias2

proc sql;
select FirstName, LastName, State, Income, DateTime, c.AccountID
from sq.smallcustomer as c inner join
sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;

Assign an alias (or nickname) to a table in the

FROM clause by adding the keyword AS.

20
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A table alias is a temporary alternate name for a table. You specify table aliases in the FROM
clause.

3.02 Activity
Open s103a02.sas from the activities fol der a nd perform the following tasks
to perform a n i nner join on tables wi th different col umns names:
1. Exa mi ne a nd run the two queries to explore the sq.statepopulation a nd
sq.statecode tabl es. Wha t columns ca n you us e to joi n the tables?
2. Speci fy the tabl es in the FROM cl a use a nd perform a n inner join. Add
the a l i as p for the sq.statepopulation ta bl e, a nd the a lias s for the
sq.statecode tabl e.
3. Compl ete the ON expression to ma tch rows where p.Name =
s.StateCode. Hi ghl ight a nd run the query. How ma ny rows a re in the new
report?

21
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Matching Rows with a Natural Join

SELECT col-name, col-name

FROM table1 NATURAL JOIN table2
proc sql;
select *
from sq.smallcustomer as c natural join
sq.smalltransaction as t;
quit; A natural join assumes
that you want to base
the join on all pairs of
common columns.

24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A natural join selects rows from two tables that have equal values in columns that share the same
name and the same type. A natural join is requested with the syntax NATURAL JOIN. If like columns
are not found, then a Cartesian product is performed.

Do not use an ON clause with a natural join. When using a natural join, an ON clause is implied,
matching all like columns. You can use a WHERE clause to subset the query results.

A natural join can be an inner join or an outer join, which is requested with the syntax INNER or
OUTER. If the join type specification is omitted, then an inner join is implied.

FEEDBACK Option

PROC SQL FEEDBACK;

proc sql feedback;

select *
from sq.smallcustomer as c natural join
sq.smalltransaction as t;
quit; The FEEDBACK option expands a SELECT statement
to the SAS log.

NOTE: Statement transforms to:

select COALESCE(T.AccountID, C.AccountID) as AccountID,

COALESCE(T.BankID, C.BankID) as BankID, T.DateTime...

25
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Selecting Data from More Than Two Tables

Results

How can I retrieve

the merchant name Merchant Name Bank Name
and the bank name
in the results?

26
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.03 Activity
Open s103a03.sas from the activities fol der a nd perform the following tasks
to fi nd tabl es that contain BankID a nd MerchantID col umns:
1. Compl ete the fi rst query by a ddi ng the BANKID col umn na me i n the
WHERE cl a use. How ma ny tables contain the BankID col umn?
2. Repl ace BANKID wi th MERCHANTID. How ma ny tabl es contain the
MerchantID col umn?

27
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Performing an Inner Join with Four Tables

Scenario

Use PROC SQL to perform an inner join among four tables.

Files
• s103d02.sas
• smallcustomer – a SAS table that contains one row per customer
• smalltransaction – a SAS table that contains one row per customer transaction
• bank – a SAS table that contains bank information
• merchant – a SAS table that contains merchant information

Syntax

PROC SQL;
SELECT col-name, col-name
FROM table1 INNER JOIN table2
ON table1.col-name=table2.col-name INNER JOIN
table3
ON join-criteria INNER JOIN
table4
ON join-criteria;
QUIT;

Notes
• To join more than two tables, each join must be listed individually in the query.

Demo

1. Open the s103d02.sas program in the demos folder and find the Demo section of the program.
Under the Explore the Tables section, run the queries to explore the sq.smallcustomer,
sq.smalltransaction, sq.bank, and sq.merchant tables. Describe the relationships between
the tables.
proc sql inobs=5;
title "Table: SMALLCUSTOMER";
select *
from sq.smallcustomer;
title "Table: SMALLTRANSACTION";
select *
from sq.smalltransaction;
title "Table: MERCHANT";

select *
from sq.merchant;
title "Table: BANK";
select *
from sq.bank;
title;
quit;
2. Find the Joining Data from More Than Two Tables section. Highlight and run the query to join
sq.smallcustomer with sq.transaction. Examine the results.
proc sql;
select FirstName, LastName, c.State, Income, DateTime, MerchantID,
Amount, c.AccountID, c.BankID
from sq.smallcustomer as c inner join
sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;
3. Add a second inner join and join the MerchantID column from the sq.merchant table with the
MerchantID column of the previous join. Replace MerchantID in the SELECT clause with
MerchantName. Highlight and run the query. Examine the results.
proc sql;
select FirstName, LastName, c.State, Income, DateTime, MerchantID
MerchantName, Amount, c.AccountID, c.BankID
from sq.smallcustomer as c inner join sq.smalltransaction as t
on c.AccountID = t.AccountID inner join
sq.merchant as m
on t.MerchantID = m.MerchantID;
quit;
4. Add a third inner join and join the BankID column from the sq.bank table with the BankID
column of the previous join. Replace BankID in the SELECT clause with the bank name.
Highlight and run the query. Examine the results.
proc sql;
select FirstName, LastName, c.State, Income, DateTime,
MerchantName, Amount, c.AccountID, c.BankID, b.Name
from sq.smallcustomer as c inner join
sq.smalltransaction as t
on c.AccountID = t.AccountID inner join
sq.merchant as m
on t.MerchantID = m.MerchantID inner join
sq.bank as b
on t.BankID = b.BankID;
quit;

Effects of Missing Values on Joins

smallcustomer2 Partial

PROC SQL treats

missing values as
matches for joins.
smalltransaction2 Partial

30
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Most database products treat missing as the absence of a value (nulls). Because they do not contain
any value, they are excluded from any conditional evaluation.

PROC SQL treats missing as missing values and matches for joins. Any missing value will match
with any other missing value of the same type (character or numeric) in a join. This could return
unexpected results.

Effects of Missing Values on Joins

Missing values are joined, and this

typically is not the desired result.
31
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Effects of Missing Values on Joins

proc sql;
select *
from sq.smallcustomer2 as c inner join
sq.smalltransaction2 as t
on c.AccountID = t.AccountID and
c.AccountID is not null;
quit;
Adding the IS NOT NULL operator to
the ON clause prevents the missing
values from joining.

32
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Non-Equijoin
smallcustomer

What if you don’t

want to join by
taxbracket equality?

33
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Non-Equijoin

select FirstName, LastName, Income,

TaxBracket
from sq.smallcustomer as c inner join
sq.taxbracket as t
on c.Income >= t.LowIncome and
c.Income <= t.HighIncome;

Use comparison operators in the

ON clause instead of equality.

34
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.04 Activity
Open s103a04.sas from the activities fol der a nd perform the following tasks
to us e a non-equijoin.
1. Compl ete the ON cl a use to joi n on rows where cus tomer Income i s
grea ter tha n the LowIncome ra nge, a nd l ess tha n or equal to the
HighIncome ra nge us ing the BETWEEN-AND where operator.
2. Wha t ta x bra cket i s Ol ga Coms tock i n?
3. Vi ew your l og. Wha t note do you s ee?

35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Syntax Summary
SELECT col-name, col-name
FROM table1 INNER JOIN table2
ON table1.column = table2.column;
Inner Join

SELECT col-name, col-name ON table1.column < table2.column AND

FROM table1 INNER JOIN table2 table1.column > table2.column;
ON table1.column = table2.column Non-equijoin
INNER JOIN table3
ON join criteria; PROC SQL FEEDBACK;
Joining More Than Two Tables PROC SQL Options

38
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Practice

Level 1

1. Inner Join

Open s103p01.sas from the practices folder. Modify the program to generate a report that
shows the breakdown of employment and marital status for customers in New York City.

a. Add a PROC SQL step to create a table named work.nyc that combines sq.customer and
sq.maritalcode. Follow the requirements below.

1) This table should include only FirstName, LastName, Employed, and MaritalStatus.

2) Perform an inner join on the Married column in the sq.customer table and MaritalCode
column in the sq.maritalcode table.

3) Filter the Zip column for customers in the 10001 ZIP code.

Partial work.nyc Table

b. Execute the PROC FREQ step to generate the crosstabulation of MaritalStatus and
Employed.

PROC FREQ Results

c. Overall, are NYC customers likely to be employed or unemployed?

d. Does this vary across marital status?

Level 2

2. Join on Inequality

Open s103p02.sas from the practices folder. Modify the program to join the sq.customer and
sq.agegroup tables based on a customer’s year of birth.

a. Add a PROC SQL step to the top of the program to create a table named work.generation
that combines sq.customer and sq.agegroup. Follow the requirements below.

1) Select FirstName and LastName, and create a column named Year to determine the
DOB year of the customer from the sq.customer table. Select the Name column from
the sq.agegroup table.

2) The StartYear and EndYear columns in the sq.agegroup table indicate the starting and
ending years for each generation. Use these columns to perform a non-equijoin using the
calculated Year value from step 1) above.

Partial work.generation Table

b. Execute the PROC SGPLOT step below your query to generate the bar chart shown below.

c. Which age group has the most customers?

Challenge

3. Joining More Than Two Tables

Open s103p03.sas from the practices folder. Modify the program to create a table named
work.births3 that will be used to create a report showing a three-year projection of births by
Region, Division, and State. The data to create this report is stored in four tables as described
below:

a. sq.statepopulation contains

1) Region (the region code)

2) Division (the division code)

3) Name (the state code)

4) Births3 (an indicator of state-level, three-year projected births).

b. sq.regioncode contains

1) RegionCode

2) RegionName.

c. sq.divisioncode contains

1) DivisionCode

2) DivisionName.

d. sq.statecode contains

1) StateCode

2) StateName.

e. Using all inner joins, combine the four tables above such that the descriptive Region,
Division, and State names are combined with the Births3 data.

Partial work.Births3 Data

f. Execute the PROC TABULATE step to generate the required report and answer the following
questions:

1) Which division has the highest projected three-year births, and which has the lowest?

2) Which region has the highest projected three-year births, and which has the lowest?

Partial PROC TABULATE Output

3.3 Outer Joins

SQL Outer Joins

Matches and Matches and all Matches and

nonmatches nonmatches nonmatches
Left Join Full Join Right Join

41
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Left joins return all rows from the left table and matching rows from the right table.

Right joins return all rows from the right table and the matching rows from the left.

Full joins return all matching and nonmatching rows from both tables.

SQL Outer Join Syntax

Join Type

SELECT columns
FROM table1 LEFT | RIGHT | FULL JOIN table2
ON table1.column = table2.column;

Left Table Right Table

42
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Scenario
smallcustomer Partial Report of all
Left Join customers with
or without a
transaction

smalltransaction Partial

43
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Left Join Syntax

Left Join select *

from sq.smallcustomer as c left join
sq.smalltransaction as t
on c.AccountID = t.AccountID;

Report of all
customers with or
without transactions

44
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A left outer join lists matching rows and rows from the left-hand table (the first table listed in the
FROM clause) that do not match any row in the right-hand table. A left join is specified with the
keywords LEFT JOIN and ON.

Scenario
smallcustomer Partial
Right Join
Report of all
transactions
with or without
a customer

smalltransaction Partial

45
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Right Join Syntax

select * Right Join

from sq.smallcustomer as c right join
sq.smalltransaction as t
on c.AccountID = t.AccountID;

Report of all transactions with or

without a customer

46
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A right join, specified with the keywords RIGHT JOIN and ON, is the opposite of a left join:
nonmatching rows from the right-hand table (the second table listed in the FROM clause) are
included with all matching rows in the output. This example reverses the join of the last example. It
uses a right join to select all the transactions from the smalltransaction table and displays the
customers only if the customer matches.

Selecting Columns in Inner Joins

c.AccountID t.AccountID

proc sql;
select FirstName, LastName, Income, AccountID,
DateTime, MerchantID, Amount
from sq.smallcustomer as c inner join
sq.smalltransaction as t With an inner join,
on c.AccountID = t.AccountID; you can select either
quit; AccountID column.

47
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Selecting Columns in Outer Joins

c.AccountID = t.AccountID

proc sql;
select FirstName, LastName, Income, AccountID,
DateTime, MerchantID, Amount
from sq.smallcustomer as c left join Depending on which
sq.smalltransaction as t AccountID column
on c.AccountID = t.AccountID; we choose, our
quit; results differ.

48
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.05 Activity
Open s103a05.sas from the activities fol der a nd perform the following tasks
to perform a l eft joi n:
1. Run the query to crea te a l eft join between the sq.smallcustomer a nd
sq.smalltransaction ta bl es. Notice the difference wi thin the AccountID
col umns in rows 8 a nd 9.
2. Remove the col umn t.AccountID i n the SELECT cl a us e. Run the query and
exa mi ne the results . How ma ny mi ssing AccountID va l ues a re i n the
res ul ts?
3. Repl ace c.AccountID wi th t.AccountID. Repl a ce the c i n the col umn l abel
wi th a t. How ma ny mi s sing AccountID va l ues are i n the results?

49
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Scenario
smallcustomer Partial
Full Join
Report of all
customers and
transactions.

smalltransaction Partial

52
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Full Join Syntax

Full Join
select *
from sq.smallcustomer as c full join
sq.smalltransaction as t
on c.AccountID = t.AccountID;

Report of all customers

and transactions

53
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

SQL Full Join Syntax

Notice that the AccountID values from

each table are not always equal.

54
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

COALESCE Function

COALESCE(argument-1, argument-2<, …argument-n>)

select coalesce(c.AccountID, t.AccountID) as AccountID

The COALESCE function returns the value of

c.A ccountID t.A ccountID
the first nonmissing argument.
. 1010183063
4040164206 .
3030165207 3030165207

55
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You can use the COALESCE function to overlay columns.

Performing a Full Join with PROC SQL

Scenario

Use PROC SQL to perform a full join between two tables.

Files
• s103d03.sas
• smallcustomer – a SAS table that contains one row per customer
• smalltransaction – a SAS table that contains customer transaction data

Syntax

PROC SQL;
SELECT col-name, col-name
FROM table1 FULL JOIN table2
ON table1.col-name=table2.col-name;
QUIT;

COALESCE(argument-1, argument-2<, …argument-n>)

Notes
• The two tables to be joined are listed in the FROM clause separated by FULL JOIN.
• A full join lists all values in both tables, with or without a match.
• The COALESCE function returns the value of the first nonmissing argument.

Demo

1. Open the s103d03.sas program in the demos folder and find the Demo section of the program.
Highlight and run the query. Examine the results. Discuss the values in both AccountID
columns.
proc sql;
select FirstName, LastName, Income, c.AccountID, t.AccountID,
DateTime, MerchantID, Amount
from sq.smallcustomer as c full join sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;

2. Modify the SELECT clause and remove the t.AccountID column. Highlight and run the query.
Examine the results. Discuss the missing values in the c.AccountID column.
proc sql;
select FirstName, LastName, Income, c.AccountID, t.AccountID,
DateTime, MerchantID, Amount
from sq.smallcustomer as c full join
sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;
3. Modify the SELECT clause and replace c.AccountID with t.AccountID. Highlight and run the
query. Examine the results. Discuss the missing values in the t.AccountID column.
proc sql;
select FirstName, LastName, Income, t.AccountID,
DateTime, MerchantID, Amount
from sq.smallcustomer as c full join
sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;
4. Modify the SELECT clause, use the COALESCE function, and add c.AccountID and
t.AccountID as arguments. Add the alias AccountID and the FORMAT=10. column modifier to
the newly created column. Highlight and run the query. Examine the results.
proc sql;
select FirstName, LastName, Income,
coalesce(c.AccountID,t.AccountID) as AccountID format=10.,
DateTime, MerchantID, Amount
from sq.smallcustomer as c full join
sq.smalltransaction as t
on c.AccountID = t.AccountID;
quit;

Identifying Nonmatching Rows

select FirstName, LastName, Income,
c.AccountID "c.AccountID",
t.AccountID "t.AccoundID",
DateTime, MerchantID
from sq.smallcustomer as c left join
sq.smalltransaction as t
on c.AccountID = t.AccountID; Produce a l i st of
customers who don't
ha ve a tra ns action.

57
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Identifying Nonmatching Rows

customers who do not have

a transaction

58
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Identifying Nonmatching Rows

The WHERE clause filters for all customers

with a missing transaction AccountID.

59
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.06 Activity
Open s103a06.sas from the activities fol der a nd perform the following tasks
to fi nd a l l tra nsactions not a ssoci ated wi th a documented cus tomer:
1. Run the query to crea te a l eft join between the sq.smalltransaction2 a nd
sq.smallcustomer2 tabl es. Exa mine the report. Notice that the rows with
mi s s ing va lues i n AccountID ha ve been joined.
2. In the ON cl a us e, a dd the expression AND t.AccountID is not null. Run
the query. Confi rm tha t mi s sing va lues were not joined.
3. Add a WHERE cl a use with the expression c.AccountID is NULL to fi l ter for
a l l tra nsactions wi thout a documented customer. Run the query a nd
exa mi ne the report. How ma ny tra nsacti ons do not have a customer
a s s oci ated wi th them?

60
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Syntax Summary

SELECT columns
FROM table1 <LEFT |RIGHT |FULL > JOIN table2
ON table1.column = table2.column
Outer Join

COALESCE(argument-1, argument-2<, …argument-n>)

COALESCE Function

64
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Practice

Level 1

4. Using Outer Joins to Find Nonmatches

Join the sq.globalpop and sq.globalmetadata tables to create the work.meta table. Use the
work.meta table to generate a report showing the country codes for countries in the
sq.globalpop table that do not have any country metadata in the sq.globalmetadata table.

a. Create a table from the join. Name the table work.meta.

b. Select the CountryCode, SeriesName, EstYear1, and EstYear3 columns from the
sq.globalpop table and the ShortName and IncomeGroup columns from the
sq.globalmetadata table.

c. Perform a left join on the sq.globalpop and sq.globalmetadata tables.

d. Use the CountryCode column in both tables for the join criteria.

Partial work.meta Table

e. Create a report showing the unique country codes for which there is no global metadata
using the work.meta table.

1) Select the CountryCode column from the work.meta table and eliminate duplicate
values.

2) Filter for rows where the ShortName column is missing. The ShortName column
contains values from the sq.globalmetadata table. If the results are missing, then the
row did not retrieve information from sq.globalmetadata.

3) Order the results by CountryCode.

f. What is the last CountryCode value in your results?

Partial Results

Level 2

5. Using Outer Joins to Summarize Data

Generate a report showing the count of customer marital status descriptions for each primary
bank. The sq.customer table contains a marital code (Married) and a primary bank ID. The final
results should contain BankID, MaritalStatus, Name (name of bank), and Count.

a. Write a PROC SQL step to join the sq.customer table and the sq.maritalcode table.

1) Select BankID from the sq.customer table and MaritalStatus value from the
sq.maritalcode table. Create a new column named Count to count the number of
customers. Format the new column using commas.
2) Use a left join to select all customers from the sq.customer table, with or without
matches in the sq.maritalcode table.

3) Use the Married column in the sq.customer table and the MaritalCode column in the
sq.maritalcode table as the join criteria.

4) Filter rows where the customer's bank ID is not missing.

5) Group the data by BankID and MaritalStatus.

6) Order the table by descending Count.

7) As a checkpoint, run the query.

Partial Results

b. In the same PROC SQL step, add the descriptive bank Name column to the results.

1) After the MaritalStatus column in the SELECT clause, add Name from the sq.bank
table based on matching BankID values, again using a left join.

2) Add Name in the GROUP BY clause after the MaritalStatus column.

3) Correctly reference the BankID columns in the query.

4) Add an appropriate title.

Partial Results

c. Which combination of MaritalStatus and Name had the lowest count of customers?

Challenge

6. Combining Inner and Outer Joins

a. Some of our New York City merchants are having difficulty with sales. Your job is to identify
merchants who have had no recent sales, capture what industry they are in, and then
generate a list of customers in their area who are transacting with other vendors to help them
market the appropriate customers to generate sales.

1) Use a left outer join to combine the sq.merchant and sq.transaction tables to identify
New York City merchants (Zip=10001) with no transactions. From the sq.merchant
table, select MerchantName, MerchantID, Type, and Zip. Label Type as Merchant
Type and Zip as Merchant Zipcode. Merchants who do not have any transactions will
be those who have no entries in the sq.transaction table.

2) As a checkpoint, the NYC merchants with no transactions are shown below.

Results

b. The next step is to find the customers in NYC who have similar Type transactions with other
vendors. The problem is that the sq.transactionfull table (where transaction information is
combined with customer information) does not contain a column for the customer’s ZIP code
for us to join on, only the full address. Therefore, we need to extract the ZIP code from the
customer.Address column to perform the final join.

1) In the same PROC SQL step, use an inner join to combine the above list with the
sq.transactionfull table by Type and Zip to generate a report with customer contact
information for all NYC customers with similar Type transactions.

Hint: Assign the sq.transactionfull table an alias of c and use the following expression
to extract Zip from the c.address column to perform this join:
input(scan(c.address,-1),5.)
2) Include CustomerID, CustomerName, and Address from the sq.transactionfull table.
Label Address as Customer Address.

3) Ensure that the report shows unique rows of data and is ordered by merchantID and
CustomerID. Add an appropriate title to the report.

Partial PROC SQL Output

c. How many potential new customers can the merchant Miasma Mitigation, Inc. now market?

3.4 Complex Joins

Reflexive Join

The employee table includes a

employee list of all employees.
EmployeeID
EmployeeName
ManagerID Fi nd ManagerName
… for ea ch empl oyee.
EmployeeID EmployeeName ManagerID ManagerName
121044 Abbott, Ray 121144
120145 Aisbitt, Sandy 120103
120761 Akinfolarin, Tameaka 120746
121144 Capachietti, Renee 121142

67
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Reflexive Join

EmployeeID EmployeeName ManagerID ManagerName

121044 Abbott, Ray 121144
120145 Aisbitt, Sandy 120103
Self-join on the
120761 Akinfolarin, Tameaka 120746
employee tabl e to
121144 Capachietti, Renee 121142
retri eve ma nager
na mes .
EmployeeID EmployeeName ManagerID ManagerName
121044 Abbott, Ray 121144 Capachietti, Renee
120145 Aisbitt, Sandy 120103
120761 Akinfolarin, Tameaka 120746
121144 Capachietti, Renee 121142

68
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Required Table Aliases

FROM table1 <AS> alias1 JOIN-TYPE

table1 <AS> alias2

select e.EmployeeID, e.EmployeeName,

e.StartDate format=date9.,
e.ManagerID, To rea d the s a me
m.EmployeeName as ManagerName tabl e twice, l i st i t
from sq.employee as e inner join twice i n the
sq.employee as m
on e.ManagerID = m.EmployeeID; FROM cl a us e.

69
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Performing a Reflexive Join

Scenario

Use PROC SQL to perform a reflexive join using a single table.

Files
• s103d04.sas
• employee – a SAS table that contains one row per employee

Syntax

PROC SQL;
SELECT col-name, col-name
FROM table1 INNER JOIN table1
ON table1.col-name=table1.col-name;
QUIT;

Notes
• A reflexive join or self-join is used to join a table to itself as if the table were two tables.
• The ON expression indicates how rows should be matched. The column names must be qualified
as table-name.col-name.

Demo

1. Open the s103d04.sas program in the demos folder and find the Demo section of the program.
Highlight and run the query. Examine the results.
proc sql;
select e.EmployeeID, e.EmployeeName, e.StartDate format=date9.,
e.ManagerID
from sq.employee as e;
quit;
2. Modify the query to create a reflexive join. In the FROM clause, add an inner join followed by the
sq.employee table again. Add the alias m to the second sq.employee table. Add the ON clause
and set e.ManagerID equal to m.EmployeeID.
proc sql;
select e.EmployeeID, e.EmployeeName, e.StartDate format=date9.,
e.ManagerID
from sq.employee as e inner join
sq.employee as m
on e.ManagerID = m.EmployeeID;
quit;

3. Add the EmployeeName column in the SELECT clause. Qualify the new EmployeeName
column with the table alias m. Highlight and run the query. Examine the results.
proc sql;
select e.EmployeeID, e.EmployeeName, e.StartDate format=date9.,
e.ManagerID, m.EmployeeName
from sq.employee as e inner join
sq.employee as m
on e.ManagerID = m.EmployeeID;
quit;
4. Add the column alias ManagerName to the m.EmployeeName column and an ORDER BY
clause to sort by ManagerName. Highlight and run the query. Examine the results.
proc sql;
select e.EmployeeID, e.EmployeeName, e.StartDate format=date9.,
e.ManagerID, m.EmployeeName as ManagerName
from sq.employee as e inner join
sq.employee as m
on e.ManagerID = m.EmployeeID
order by ManagerName;
quit;

Scenario
transactionfull statecode

How would you

join these tables
to retrieve
StateName?

71
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Functions to Join Tables

transactionfull statecode

Use the SUBSTR

function to extract
the first two
characters from
StateID.

select StateID, CustomerName, StateName

from sq.transactionfull as t inner join
sq.statecode as s
on substr(t.StateID,1,2) = s.StateCode;

72
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.07 Activity
Open s103a07.sas from the activities fol der a nd perform the following tasks
to joi n two tabl es us ing the SUBSTR function:
1. Exa mi ne a nd run the query. Di d you recei ve a s yntax error?
2. In the ON cl a us e, us e the SUBSTR function on t.StateID to extra ct the
first two characters. Run the query. Whi ch StateName i s Caberto, Glen
Daniel from?

73
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Functions to Join Tables

customerzip

Can the columns you

use to join tables have a
different column type?
Character ZipCode
sashelp.zipcode

Numeric ZIP

75
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.08 Activity
Open s103a08.sas from the activities fol der a nd perform the following tasks
to joi n a numeric col umn wi th a character col umn:
1. Run the queri es i n the Create a Table and Insert Values s ection. View the
newl y crea ted table.
2. Run the query i n the Join Different Column Types s ection. What s yntax
error wa s generated when you joi ned columns of di fferent types ?
3. Lea ve the progra m open for the next a cti vi ty.

76
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Converting Column Value Functions

character arithmetic
function calculation

Function What it does

INPUT(source, informat) Converts character values to numeric values using a

specified informat
PUT(source, format) Converts numeric or character values to character
values using a specified format

78
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For more information about the PUT function, see "PUT Function" in the SAS® 9.4 Functions and
CALL Routines: Reference, Fifth Edition documentation. You can also find the direct link in the
Course Links section on the ELP.

For more information about the INPUT function, see "INPUT Function" in the SAS® 9.4 Functions
and CALL Routines: Reference, Fifth Edition documentation. You can also find the direct link in the
Course Links section on the ELP.

Converting Numeric to Character Values

put(z.Zip,z5.)

The Z format writes

standard numeric
data with leading 0s. The PUT function with the Z format converts
the numeric ZIP code 4429 to 04429.

79
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.09 Activity
Us e s103a08.sas from the previ ous a ctivi ty a nd perform the following ta sks
to joi n a numeric col umn wi th a character col umn by us i ng the PUT
function.
1. If you ha ve not run the queri es i n the Create a Table and Insert Values
s ecti on, run those now.
2. Us e the PUT functi on to convert z.Zip i n the ON cl a use to a cha ra cter
va l ue using the z5 forma t. Run the query.
3. Wha t ci ty does the ZipCode va l ue 14216 repres ent?

80
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Beyond SQL Essentials

What if you want to ...

. . . l ea rn more a bout . . . downl oad the SQL . . . l ea rn more a bout

mergi ng us ing the Joi n Summa ry chea t functi ons i n PROC
DATA s tep? s heet? SQL?

• Take the SAS Programming 2 • Visit the Course Handouts • Read the SAS paper Top 10
course to learn more about section on the ELP and Most Powerful Functions for
DATA step match-merges. download the SQL Join PROC SQL.
• Visit the SQL and the DATA Summary PDF.
Step section on the ELP for
additional resources about
comparisons of the DATA
step and PROC SQL.

82
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.5 Solutions
Solutions to Practices
1. Inner Join
/*s103s01.sas*/

proc sql;
create table work.NYC as
select c.FirstName, c.LastName, c.Employed, m.MaritalStatus
from sq.customer as c inner join
sq.maritalcode as m
on c.married=m.MaritalCode
where zip=10001;
quit;

/*Alternate Solution*/
proc sql;
create table work.NYC as
select c.FirstName, c.LastName, c.Employed, m.MaritalStatus
from sq.customer as c, sq.maritalcode as m
where c.married=m.MaritalCode and
zip=10001;
quit;

title 'Marital Status by Employment for NYC Customers';

proc freq data=work.NYC order=Freq;
tables MaritalStatus*employed;
run;
title;
Overall, are NYC customers likely to be employed or unemployed? Employed

Does this vary across marital status? Yes, NYC customers whose marital status is Single
are more likely to be unemployed.

2. Join on Inequality
/*s103s02.sas*/

/*Solution*/
proc sql;
create table work.generation as
select c.FirstName,
c.LastName,
year(c.DOB) as Year,
a.Name
from sq.Customer as c inner join
sq.AgeGroup as a

on year(c.DOB) between a.StartYear and a.EndYear;

quit;

/*Alternate Solution*/
proc sql;
create table work.generation as
select c.FirstName,
c.LastName,
year(c.DOB) as Year,
a.Name
from sq.Customer as c, sq.AgeGroup as a
where calculated Year between a.StartYear and a.EndYear;
quit;

/*Bar Chart*/
title 'Count of Customers by Generation';
proc sgplot data=work.generation noautolegend;
hbar Name /
stat=freq
dataskin=sheen
categoryorder=respdesc
datalabel
datalabelattrs=(size=9pt)
FILLATTRS=(color=cx6f7eb3);
yaxis label='Generation';
xaxis grid label='Count';
run;
title;
Which age group has the most customers? Millennials or Generation Y

3. Joining More Than Two Tables

/*s103s03.sas*/

proc sql;
create table Births3 as
select c.RegionName, d.DivisionName, b.Statename, a.Births3
from sq.statepopulation as a inner join
sq.statecode as b
on a.name=b.statecode inner join
sq.RegionCode as c
on a.region=c.regionCode inner join
sq.divisioncode as d
on a.Division=d.divisioncode;
quit;

/*Alternate Solution*/
proc sql;
create table Births3 as

select c.RegionName, d.DivisionName, b.Statename, a.Births3

from sq.statepopulation as a,
sq.statecode as b,
sq.RegionCode as c,
sq.divisioncode as d
where a.name=b.statecode and
a.region=c.regionCode and
a.Division=d.divisioncode;
quit;

title 'Projected 3-Year Births by Region/Division/State';

proc tabulate data=WORK.BIRTHS3;
var Births3;
class RegionName / order=unformatted missing;
class DivisionName / order=unformatted missing;
class StateName / order=unformatted missing;
table
Sum={label=""}*(
RegionName={label= 'Region'}*(
DivisionName={label= 'Division'}*(
StateName={label="State"}
all={label= 'Division Total'}*{style={BACKGROUND=#FFFF99}})
all={label= 'Region Total'}*{style={BACKGROUND=#CCFFFF}})
all={label= 'Overall Total'}*{style={BACKGROUND=#CCFFCC}}),
Births3={label= 'Projected 3-Year Births'}*F=comma12.;
run;
Which division has the highest projected three-year births, and which has the lowest?
South Atlantic has the highest. New England has the lowest.

Which region has the highest projected three-year births, and which has the lowest?
South region has the highest. Northeast region has the lowest.

4. Using Outer Joins to Find Nonmatches

/*s103s04.sas*/

************************************;
*Solution with Temporary Table *;
************************************;
proc sql;
create table work.meta as
select p.CountryCode, p.SeriesName, p.EstYear1, p.EstYear3,
m.ShortName, m.IncomeGroup
from sq.globalpop as p left join
sq.globalmetadata as m
on p.countrycode= m.countrycode;
quit;

title 'Countries with no Metadata';

proc sql;

select distinct CountryCode

from work.meta
where ShortName is null
order by CountryCode;
quit;
title;

************************************;
*Solution with No Temporary Table *;
************************************;
title 'Countries with no Metadata';
title2 'No Temporary Table';
proc sql;
select distinct p.CountryCode
from sq.globalpop as p left join
sq.globalmetadata as m
on p.CountryCode= m.CountryCode
where ShortName is null;
quit;
title;
What is the last CountryCode value in your results? WSM

5. Using Outer Joins to Summarize Data

/*s103s05.sas*/

/*a*/
proc sql;
select c.BankID,
m.MaritalStatus,
count(*) as Count format=comma10.
from sq.customer as c left join
sq.maritalcode as m
on c.Married=m.MaritalCode
where c.BankID is not null
group by c.BankID, m.MaritalStatus
order by Count desc;
quit;

/*b*/
title 'Count of Marital Status by Bank';
proc sql;
select c.BankID,
m.MaritalStatus,
b.Name,
count(*) as Count format=comma10.
from sq.customer as c left join
sq.maritalcode as m
on c.Married=m.MaritalCode left join
sq.bank as b

on c.BankID=b.BankID
where c.BankID is not null
group by c.BankID, m.MaritalStatus, b.Name
order by Count desc;
quit;
title;
Which combination of MaritalStatus and Name had the lowest count of customers?
Widowed and Wheatberry Bank, Inc. had a total count of 362.

6. Combining Inner and Outer Joins

/*s103s06.sas*/

/*a*/
title 'NYC Merchants with no Transactions';
proc sql;
select a.MerchantName,
a.MerchantID,
a.Type 'Merchant Type',
a.Zip 'Merchant Zipcode'
from sq.merchant as a left join
sq.transaction as b
on a.MerchantID=b.MerchantID
where a.Zip=10001 and b.MerchantID is null;
quit;
title;

/*b*/
proc sql;
title 'Customers with similar Type Transactions';
title2 'as NYC Merchants with no Sales';
select distinct a.MerchantName,
a.MerchantID,
a.type 'Merchant Type',
a.Zip 'Merchant Zipcode',
c.CustomerID,
c.Customername,
c.Address 'Customer Address'
/*Join merchants with transactions. Notice where clause below
that determines which rows are returned.*/
from sq.merchant as a left join
sq.transaction as b
on a.MerchantID=b.MerchantID
/*Combine the above list with all customers with similar type
transactions.*/
inner join sq.transactionfull as c
on a.type=c.type and input(scan(c.address,-1),5.)=a.zip
/*the where filter selects NYC merchants who do not exist in
the transaction table=no sales*/
where a.zip=10001 and b.MerchantID is null

order by 2,5;
quit;
title;
How many potential new customers can the merchant Miasma Mitigation, Inc. now market?
Six potential new customers

Solutions to Activities and Questions

continued...
3.01 Activity – Correct Answer
2. Wha t note do you s ee?

NOTE: The execution of this query involves performing one or

more Cartesian product joins that can not be optimized.

The default JOIN combines every row in each table. This is

called the Cartesian product. Typically, the Cartesian
product is not the desired result.

9
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.01 Activity – Correct Answer

Redundant
Partial columns

Nonmatching IDs

NOTE: The execution of this query involves performing one or more

Cartesian product joins that can not be optimized.

10
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
3.02 Activity – Correct Answer
1. Wha t col umns ca n you us e to joi n the ta bles? The Name column from
the statepopulation table and StateCode column from the statecode
table

22
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.02 Activity – Correct Answer

3. Compl ete the ON expression to ma tch rows when p.Name =
s.StateCode. Hi ghl ight a nd run the query. How ma ny rows a re in the new
report? 52

select Name, StateName, PopEstimate1, PopEstimate2,

PopEstimate3
from sq.statepopulation as p inner join In SQL, join
sq.statecode as s condition columns
on p.Name = s.StateCode
order by StateName; do not need to
have the same
names.

23
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.03 Activity – Correct Answer

1. How ma ny ta bl es contain the BankID col umn? Eight

contains the full BANK

information

2. How ma ny tabl es contain the MerchantID col umn? Five

contains the full MERCHANT

information

28
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
3.04 Activity – Correct Answer
1.
select FirstName, LastName,
Income format=dollar16., TaxBracket
from sq.smallcustomer as c inner join
sq.taxbracket as t
on c.Income between t.LowIncome and t.HighIncome
order by TaxBracket desc, Income desc;

Use the BETWEEN-AND special

where operator.

2. Wha t tax bra cket i s Ol ga Coms tock i n? 12%

36
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.04 Activity – Correct Answer

3. Wha t note do you s ee?

select FirstName, LastName,

Income format=dollar16., TaxBracket
from sq.smallcustomer as c inner join
sq.taxbracket as t
on c.Income between t.LowIncome and t.HighIncome
order by TaxBracket desc, Income desc;

NOTE: The execution of this query involves performing one

or more Cartesian product joins that can not be
optimized.

37
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
3.05 Activity – Correct Answer
1. Noti ce the di fference wi thin the AccountID col umns i n rows 8 a nd 9.

The left join selects all customers

with or without a transaction.
50
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.05 Activity – Correct Answer

No missing AccountID values Two missing AccountID values

selects AccountID values from selects AccountID values from

the smallcustomer table the smalltransaction table
51
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
3.06 Activity – Correct Answer
1. Noti ce tha t the rows wi th mi ssing va lues i n AccountID ha ve been joined.

Missing values rows were

joined.

61
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
3.06 Activity – Correct Answer
2. Confi rm tha t mi s sing va l ues were not joined.

on c.AccountID = t.AccountID and t.AccountID is not null;

Do not join missing values.

62
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.06 Activity – Correct Answer

3. How ma ny tra ns a ctions do not have a customer a ssociated wi th them?
seven transactions

on c.AccountID = t.AccountID and t.AccountID is not missing

where c.AccountID is null;

Find all transactions without

a documented customer.

63
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.07 Activity – Correct Answer

1. Di d you recei ve a syntax error?
No syntax error, but no rows were returned from the query.
NOTE: No rows were selected.

2. Whi ch StateName i s Caberto, Glen Daniel from? California

select StateID, CustomerName, StateName
from sq.transactionfull as t inner join
sq.statecode as s
on substr(t.StateID,1,2) = s.StateCode;

74
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.08 Activity – Correct Answer

2. Wha t s ynta x error wa s generated when you joi ned col umns of
di fferent types?

select c.CustomerID, c.ZipCode, c.Gender,

z.Zip, z.City, z.StateCode
from customerzip as c inner join
sashelp.zipcode as z
on c.ZipCode = z.Zip;

Character ZIP code Numeric ZIP code

ERROR: Expression using equals (=) has components that are of different data types.

77
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

3.09 Activity – Correct Answer

3. Wha t ci ty does the ZipCode va l ue 14216 repres ent? Buffalo
select c.CustomerID, c.ZipCode, c.Gender,
z.Zip, z.City, z.StateCode
from customerzip as c inner join
sashelp.zipcode as z
on c.ZipCode = put(z.Zip,z5.);
quit; The Zip column
converts to character
in the ON clause.

81
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
Lesson 4 Subqueries
4.1 Subquery in WHERE and HAVING Clause s .................................................................. 4-3
Demonstration: Subquery That Returns a Single Value................................................. 4-7
Demonstration: Subquery That Returns Multiple Values.............................................. 4-12
Practice............................................................................................................... 4-19

4.2 In-Line Views (Query in the FROM Clause)................................................................. 4-25

Demonstration: Using an In-Line View ..................................................................... 4-29
Practice............................................................................................................... 4-39

4.3 Subquery in the SELECT Clause ................................................................................ 4-43

Demonstration: Remerging Summary Statistics......................................................... 4-46
Practice............................................................................................................... 4-51

4.4 Solutions ................................................................................................................... 4-54

Solutions to Practices ............................................................................................ 4-54
Solutions to Activities and Questions........................................................................ 4-61
4-2 Lesson 4 Subqueries

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-3

4.1 Subquery in WHERE and HAVING

Clauses

Subquery in the WHERE and HAVING Clauses

Outer Query
SELECT col-name, col-name
FROM input-table
WHERE column operator (SELECT col-name
FROM input-table…);
Subquery
(select avg(PopEstimate1)
from sq.statepopulation
where … )…

3
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A subquery in the WHERE and HAVING clauses

• returns values to be used in the outer query’s WHERE or HAVING clause
• must return only a single column
• can return multiple values or a single value.

Noncorrelated Subqueries

Outer Query
SELECT … Independent
FROM … Subquery
<WHERE …> (select avg(PopEstimate1)
<GROUP BY …> from sq.statepopulation
where … )…
<HAVING …>
<ORDER BY …>;

A noncorrelated subquery is a self-contained query.

It executes independently of the outer query.

4
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

This course focuses on noncorrelated subqueries and refers to them as simply subqueries for ease.

Scenario

State PopEstimate1 > Total avg(PopEstimate1)

Create a report that displays

states with PopEstimate1
values greater than the
average PopEstimate1 value
of all states.

5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-5

Subqueries Steps

Calculate the subquery’s value.

1 statepopulation
Name PopEstimate1
AL 4864745
Use the value from the subquery AK 741504
2 in the outer query. AZ 6945452
AR 2990410
CA 39209127
CO 5540921
3 Combine the subquery with
the outer query.

6
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Solution without a Subquery

proc sql;
1
select avg(PopEstimate1) as Average
from sq.statepopulation;
quit;

proc sql;
2 What happens if
PopEstimate1
select Name, PopEstimate1
from sq.statepopulation changes in the data?
where PopEstimate1 > 6278420;
quit;

7
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery That Returns a Single Value

3
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > (select avg(PopEstimate1)
from sq.statepopulation);

The subquery
The subquery is evaluated first. executes
independently of the
outer query.

8
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery That Returns a Single Value

The outer query uses the value

returned by the subquery.

3
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > (6278420);

9
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-7

Subquery That Returns a Single Value

Scenario
Use a noncorrelated subquery that returns a single value.

Files
• s104d01.sas
• statepopulation – a SAS table that contains estimated state populations for the next three years

Syntax

PROC SQL;
SELECT col-name, col-name
FROM input-table
WHERE column operator (SELECT col-name
FROM input-table…)
ORDER BY col-name <DESC>;
QUIT;

Notes
• A subquery is a query expression that is nested as part of another query expression.
• A single-value subquery returns a single row and column.

Demo
1. Open the s104d01.sas program in the demos folder and find the Demo section. Run the query
to explore the sq.statepopulation table.
2. Run the query in the Future Subquery section to find the average of PopEstimate1 in the
sq.statepopulation table. View the results.
proc sql;
select avg(PopEstimate1)
from sq.statepopulation;
quit;
3. Complete the outer query to find all states that have a higher PopEstimate1 value than the
result of the previous query's value of 6278420. Sort the results by descending PopEstimate1.
proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > 6278420
order by PopEstimate1 desc;
quit;

4. Remove the value 6278420 in the WHERE clause, replace it with the query that found the
average of PopEstimate1, and enclose the inner query in parentheses.
Note: Remember to remove the semicolon from the subquery when copying and pasting.
proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > (select avg(PopEstimate1)
from sq.statepopulation)
order by PopEstimate1 desc;
quit;
5. Replace the subquery with a new query. Find the average of Population_2010 from the
sashelp.us_data table.
Note: The table in a subquery can be a different table than the outer query.
proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > (select avg(Population_2010)
from sashelp.us_data)
order by PopEstimate1 desc;
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-9

4.01 Activity
Open s104a01.sas from the activities fol der a nd perform the following tasks
to us e a subquery to return two col umns i n the WHERE cl a use:
1. Exa mi ne a nd run the fi rst query. Confi rm tha t the res ults contain one
row a nd two col umns.
2. Add the fi rs t query a s a subquery i n the s econd query to fi nd a ll s tates
wi th PopEstimate1 hi gher tha n the a verage estimated s tate population.
3. Run the query. Wha t i s the s yntax error i n the l og?

11
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery in the HAVING Clause

Division avg(PopEstimate1) > Total avg(PopEstimate1)

sq.statepopulation

14
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery in the HAVING Clause

1
proc sql;
select avg(PopEstimate1) as Average
from sq.statepopulation;
quit;

2
proc sql;
select Division, avg(PopEstimate1) as avgDivisionPop
from sq.statepopulation
group by Division
having avgDivisionPop > 6278420;
quit;

15
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.02 Activity
Open s104a02.sas from the activities fol der a nd perform the following tasks
to perform a s ubquery i n the HAVING cl a use:
1. Exa mi ne a nd run the fi rst query. Vi ew the results.
2. Modi fy the s econd query. Copy the va l ue returned by the fi rst query
i nto the s ubquery a ga inst the HAVING cl ause to return di visions wi th
a n a vera ge PopEstimate1 va l ue greater tha n the total a verage of
PopEstimate1.
3. Remove the s tatic va lue a nd a dd the subquery i n the HAVING cl a use.
4. How ma ny di vi s ions have a hi gher a verage PopEstimate1 tha n the
a vera ge PopEstimate1 of a l l the s tates?

16
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-11

Subquery That Returns Multiple Values

statepopulation Results sq.customer

Create a table of all

customers who reside
in a state in Division 3.

18
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery That Returns Multiple Values

proc sql;
1
select Name
from sq.statepopulation
where Division = '3';
quit;

2
proc sql;
create table division3 as
select *
from sq.customer
where State in ("IL","IN","MI","OH","WI");
quit;
19
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery That Returns Multiple Values

Scenario
Use a noncorrelated subquery that returns multiple values.

Files
• s104d02.sas
• customer – a SAS table that contains one row per customer
• statepopulation – a SAS table that contains estimated state populations for the next three years

Syntax

PROC SQL;
SELECT col-name, col-name
FROM input-table
WHERE column IN (SELECT column
FROM input-table…)
ORDER BY col-name <DESC>;
QUIT;

Notes
• A multiple-value subquery can return more than one value from a column.
• A multiple-value subquery can be used in a WHERE or HAVING expression that contains an
IN operator.

Demo
1. Open the s104d02.sas program in the demos folder and find the Demo section.
2. Examine the Future Subquery section and run the query to find the states that reside in
Division 3. View the results.
proc sql;
select Division, Name
from sq.statepopulation
where Division = '3';
quit;
3. Using the results from the Future Subquery section, modify the outer query by entering the
State abbreviations ("IL", "IN","MI","OH","WI") in the WHERE clause. Run the query and view
the results.
proc sql;
create table Division3 as
select *
from sq.customer
where State in ("IL", "IN","MI","OH","WI");
quit;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-13

4. Replace the values in parentheses with the query from step 2 by copying and pasting the query
inside the parentheses. The query is now the subquery. Be sure to remove the semicolon from
inside the parentheses. Highlight and run the query. Discuss the error in the log.
Note: This program will return a syntax error.
…
where State in (select Division, Name
from sq.statepopulation
where Division = '3');
quit;

5. Remove the Division column from the subquery. Highlight and run the query. Discuss the
results.
…
where State in (select Division, Name
from sq.statepopulation
where Division = '3');
quit;

6. Change the Division value from 3 to 6 in the subquery. Replace the 3 at the end of the new
table name in the CREATE TABLE statement to a 6. Highlight and run the query. Discuss the
results.
proc sql;
create table division6 as
select *
from sq.customer
where State in (select Name
from sq.statepopulation
where Division = '6');
quit;

ANY Keyword

WHERE column = ANY(subquery values)

Equivalent to
proc sql; using the IN
create table Division3 as operator
select *
from sq.customer
where State = any(select Name
from sq.statepopulation
where Division = '3');
quit;

21
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Similar to the IN operator, you can also use the ANY expression for all returned values from the
subquery.

Scenario

New York
Florida
Which states have a
PopEstimate1 value
that is greater than
New York or Florida?

22
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-15

ANY Keyword

select Name, PopEstimate1

from sq.statepopulation
where PopEstimate1 > any(select PopEstimate1
from sq.statepopulation
where Name in ("NY","FL"));

20629982,19641589

The ANY keyword is true when the value of the specified column
is greater than any of the values returned by the subquery.

23
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

MIN Function

select Name, PopEstimate1

from sq.statepopulation
where PopEstimate1 > (select min(PopEstimate1)
from sq.statepopulation
where Name in ("NY","FL"));

19641589

You can also use the MIN function inside the subquery to
return the minimum PopEstimate1.

24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You can also use > or < to return values greater than any of the values or less than any of the
values.
In this example, you can also use the MIN function inside the subquery to return the minimum value
of PopEstimate1 and use a comparison operator to compare the values.

4.03 Activity
Open s104a03.sas from the activities fol der a nd perform the following tasks
to fi nd a l l s tates with a PopEstimate1 va l ue that i s l ower tha n the va lue for
New York or Fl ori da:
1. Compl ete the query us ing the ANY keyword or MAX s tatistic.
2. Run the query. How ma ny s tates have estimated populations lower tha n
New York or Fl ori da?

25
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Correlated Subqueries
Outer Query
SELECT …
FROM …
<WHERE …>
Dependent
<GROUP BY …>
<HAVING …> Correlated
<ORDER BY …>; subqueries are
Correlated resource intensive.
(SELECT …
FROM …
<WHERE …>)

27
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

The previous subqueries have been noncorrelated subqueries that are self-contained and that
execute independently of the outer query.
A correlated subquery is dependent on the outer query. A correlated subquery requires one or more
values to be passed to it by the outer query before the subquery can be resolved. This means that
PROC SQL must process the correlated subquery multiple times : once for each table row that the
outer query processes. Correlated subqueries tend to use more resources than noncorrelated
subqueries. You typically want to avoid a correlated subquery, and we will briefly discuss why.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-17

Correlated Subqueries

select count(*) as TotalCustomer

from sq.customer as c
where '1' = (select Division
from sq.statepopulation as a
where a.Name = c.State);

How many customers

The inner query needs are from a state in
information from the outer query. Division 1?

28
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A correlated subquery is not stand-alone. It needs additional information from the main query.

Correlated Subqueries
Executing a correlated subquery
can be resource intensive.
select count(*) as TotalCustomer
from sq.customer as c
where '1' = (select Division
from sq.statepopulation as s
where s.Name = c.State);

Row from Each subquery is

outer query executed once for
every row of the outer
Row included Execute query.
in results inner query

Test WHERE
condition
29
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Conceptually, correlated subqueries are pretty tricky and resource intensive. Using correlated
subqueries reduces performance in your system. A better method is joining the tables and using a
WHERE clause to obtain the necessary rows.

Using Joins
select count(*) as TotalCustomer
from sq.customer as c
where '1' = (select Division
from sq.statepopulation as s
where s.Name = c.State);

A more concise method

is to use a join.
select count(*) as TotalCustomer
from sq.customer as c inner join
sq.statepopulation as p
on p.Name = c.State
where Division = '1';

30
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subqueries can sometimes be replaced with joins. In this example, instead of the tricky and
resource-intensive correlated subquery, you can use a join to retrieve the results.
The join joins the sq.customer and sq.statepopulation tables on State abbreviations. Then the
query filters for rows where Division equals 1.

Syntax Summary
SELECT col-name, col-name
FROM input-table
WHERE column operator (SELECT col-name
FROM input-table…)
GROUP BY col-name
HAVING column operator (SELECT col-name
FROM input-table…);
WHERE and HAVING Subqueries

WHERE column IN (subquery values)

WHERE column = ANY(subquery values)
WHERE column > ANY(subquery values)
Multiple Values Returned by a Subquery
31
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-19

Practice

Level 1
1. Subquery That Returns a Single Value
The sq.statepopulation table contains estimated population statistics for every state in the US
and its territories. Which states have an estimated three-year population growth greater than the
average for all states?
a. Write a query that displays the average three-year population growth for all states. Use the
nPopChg3 column to calculate the average from the sq.statepopulation table.
Results

b. Use the query from step a to display the states that have a projected three-year population
growth greater that the overall average.
1) Include Name and nPopChg3 in the results.
2) Label nPopChg3 as Estimated Growth and format the values with commas.
3) Use the query from step a to subset the table.
4) Order the results by descending nPopChg3.
5) Add an appropriate title to the report.
Partial Results

c. Which state has the lowest estimated growth your results?

Level 2
2. Subquery with Multiple Functions
Determine which Texas customers have higher than average credit scores.
a. Using the sq.customer table, write a query to define the high-credit threshold as 2 standard
deviations greater than the mean of CreditScore for TX customers.
1) Use the following expression to create the column HighScore. This will be your future
subquery result.
sum(avg(creditscore),(2*std(creditscore))
2) Run the query and compare your results.
Results

b. Write a second query to create a report showing Texas customers whos e CreditScore value
is greater than the results from the previous query. Use the query from part a as a subquery.
1) Select CustomerID, FirstName, LastName, and CreditScore.
2) Subset the table for customers who are from the state of Texas and who have a greater
CreditScore value than the value from the query in part a.
3) Order the report by descending CreditScore.
4) Add an appropriate title.
Partial Results

c. In your report, what is the lowest value for CreditScore?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-21

3. Subquery Using Different Tables

Create a report that lists all countries and their forecasted percentage of Borrowed for health or
medical purposes (% age 15+) that reside in Europe or Central Asia and are considered high
income.
a. Using the sq.globalmetadata table, return all the country codes for the region of Europe &
Central Asia that are in the High Income group.
Partial Report

b. Write the second query using the sq.globalfindex table and view the estimated percentage
of population who borrowed for health or medical purposes (% age 15+) for the region of
Europe & Central Asia.
1) Select CountryCode and create two new columns that convert EstYear1 and EstYear3
to percentages.
• Calculate the columns by dividing EstYear1 and EstYear3 by 100. Name the new
columns EstPct1 and EstPct3 respectively and format them using the percent format.
2) Subset the table where IndicatorName is equal to Borrowed for health or medical
purposes (% age 15+), and use the previous query to subset the data by CountryCode
values in Europe and Central Asia.
Hint: You can use the UPCASE function to standardize character casing.
3) Order the report by descending EstYear1.
4) Add an appropriate title.

Partial Report

c. Which country in your report has the lowest EstPct1 value for next year?
4. Using a Subquery with Summarized Data

Determine which customers spend less on average utility bills than the overall utility bill average?
a. Use the sq.transactionfull table to create a report showing CustomerID and the average
utility payment amount, named UtilityAmt, for customers whose average utility payment is
less than the overall average utility payment for all customers.
Hint: Use the Type column to determine which transactions are utility payments .
1) Order by descending UtilityAmt.
2) Add an appropriate title to the report.
Results

b. How many customers in the sq.transaction table have lower than average utility payments?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.1 Subquery in WHERE and HAVING Clauses 4-23

Challenge
5. Using Nested Subqueries
Create a report that lists all countries that have a higher population than the average of all
countries and where the EstYear1 forecast for Outstanding housing loan (% age 15+) increases
in EstYear3.
a. Open s104p05.sas from the practices folder. Run the query to create the table
CountryEstPop. The table contains total estimated population for ages 15+ for each
country.
b. Write a query to find all countries in the CountryEstPop table with a higher population than
the mean of all countries.
Partial Results

c. Use the previous query as a subquery. Select all columns from the sq.globalfindex table
where IndicatorName is Outstanding housing loan (% age 15+), the EstYear1 forecast
increases in three years, EstYear1 and EstYear3 are not null, and the country has a
population estimate greater than the mean of all countries.
1) Convert EstYear1 and EstYear3 to a percent and format the values. Create a new
column to determine the percent increase by subtracting EstYear3 by EstYear1. Name
the column PctIncrease and format the values using a percent.
2) Order the results by descending PctIncrease. Add an appropriate title.

Results

d. Which country had the largest value for PctIncrease?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-25

4.2 In-Line Views (Query in the FROM

Clause)

In-Line View
Outer Query
SELECT …
FROM (SELECT col-name
FROM … An in-line view acts
<WHERE …>) as a virtual table.
<WHERE …>
<GROUP BY …>
<HAVING …>
<ORDER BY …>;

34
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

In-Line View

proc sql;
select *
from (select CustomerID, State, Income
from sq.customer
where Income > 100000);
quit; You cannot use an
ORDER BY clause in
an in-line view.
Creates a virtual table to use in
the outer query.

35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Here are some characteristics of in-line views:

• An in-line view is not assigned a permanent name, although it can take an alias.
• An in-line view can be referred to only in the query in which it is defined. It cannot be referenced in
another query.
• You cannot use an ORDER BY clause in an in-line view.
• The names of columns in an in-line view can be assigned in the column list of that in-line view.
This syntax can be useful for renaming columns.
• In order to visually separate an in-line view from the rest of the query, you can enclose the in-line
view in any number of pairs of parentheses. If you specify an alias for the in-line view, then the
alias specification must appear outside the outermost pair of parentheses for that in-line view.

Scenario
12.0%

Create a report that shows the

percentage of customers in each
state based on each state’s
estimated population.

36
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-27

Using Temporary Tables

create table totalcustomer as

1
select State,count(*) as TotalCustomer
from sq.customer
group by State;

Count the number of

customers in each state.

37
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Temporary Tables

select c.State, 2
c.TotalCustomer,
s.EstimateBase,
c.TotalCustomer/s.EstimateBase as
PctCustomer format=percent7.3
from totalcustomer as c inner join
sq.statepopulation as s
on c.State = s.Name
order by PctCustomer;

Join with the

sq.statepopulation table.

38
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using Temporary Tables

What if the customer

1 2 table is updated daily
and we always want the
most recent results?

Create the Join totalcustomer with

totalcustomer table. sq.statepopulation.

39
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-29

Using an In-Line View

Scenario
Use an in-line view to create a virtual table.

Files
• s104d03.sas
• customer – a SAS table that contains one row per customer
• statepopulation – a SAS table that contains estimated state populations for the next three years

Syntax

PROC SQL;
SELECT col-name, col-name
FROM (SELECT column,…
FROM input-table…)
WHERE expression
ORDER BY col-name <DESC>;
QUIT;

Notes
• An in-line view is a query in the FROM clause.
• An in-line view produces a virtual table that the outer query uses to select data.
• An in-line view can be referenced only in the query in which it is defined.

Demo
1. Open the s104d03.sas program in the demos folder and find the Demo section. Run the first
query to explore the statepopulation and customer tables.
2. Move to the next section, Temporary Table Solution. Discuss both queries in the section.
a. Run the first query to create the totalcustomer temporary table. View the results.
b. Run the second query to join the totalcustomer and sq.statepopulation tables and
calculate the new column PctCustomer that calculates the percentage of customers in each
state based on the current year's estimated population. View the results.
3. Copy only the query that is used to create the totalcustomer table. Move to the Using an In-
Line View section of the program. Paste the query (not including the CREATE TABLE
statement) in the FROM clause to create an in-line view. Be sure to remove the semicolon.
Highlight and run the query. View the syntax error in the log.

…
from (select State,count(*) as TotalCustomer
from sq.customer
group by State
order by TotalCustomer desc) as c inner join
sq.statepopulation as s
on c.State = s.Name
…

4. Remove the ORDER BY clause in the subquery. Run the query and view the results.
…
from (select State,count(*) as TotalCustomer
from sq.customer
group by State
order by TotalCustomer desc) as c inner join
sq.statepopulation as s
on c.State = s.Name
…

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-31

Storing an In-Line View

…
from (select State,count(*) as TotalCustomer
from sq.customer
group by State)
…;

How can you store the

in-line view to use in
other queries?

41
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Creating a View

select State,count(*) as TotalCustomer

from sq.customer
group by State);

• is a stored query and contains no actual data

• can be derived from one or more tables, PROC SQL view
views, DATA step views, or SAS/ACCESS views
• accesses the most current data
• can be referenced in SAS programs
• cannot have the same name as a data table stored in
the same SAS library

42
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

CREATE VIEW Statement

CREATE VIEW view-name AS
SELECT …; The query is stored as a
permanent view in the
proc sql; sq library.
create view sq.totalcustomer as
select State,count(*) as TotalCustomer
from sq.customer
group by State;
quit;

sq.totalcustomer

43
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

For sake of efficiency, it is recommended that you avoid using the ORDER BY clause in a query that
defines a view. Using the ORDER BY clause in a view definition forces PROC SQL to sort the data
every time that the view is referenced. Instead, you can use an ORDER BY clause in queries that
reference the view.

Using a VIEW
sq.totalcustomer

You can use a view

in place of the
in-line view.

select c.State, c.TotalCustomer, s.EstimateBase,

c.TotalCustomer/s.EstimateBase as
PctCustomer format=percent7.3
from (select State,count(*) as TotalCustomer
from sq.customer
group by State) as c inner join
sq.statepopulation as s
on c.State = s.Name
order by PctCustomer;

44
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-33

Using a VIEW
sq.totalcustomer
The view executes the
stored query and
extracts the most
current data.

select c.State, c.TotalCustomer, s.EstimateBase,

c.TotalCustomer/s.EstimateBase as
PctCustomer format=percent7.3
from sq.totalcustomer as c inner join
sq.statepopulation as s
on c.State = s.Name
order by PctCustomer;

45
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.04 Activity
Open s104a04.sas from the activities fol der a nd perform the following tasks
to crea te a nd us e a vi ew:
1. Crea te a vi ew na med VWtotalcustomer from the query. Run the query
a nd exa mine the log.
2. Run the code i n the s ection Use the View in the PROCS Below. Whi ch
s ta te has the hi ghest number of cus tomers?

46
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Location of a PROC SQL View

proc sql;
create view sq.totalcustomer as
select State,count(*) as TotalCustomer
from customer
group by State;
quit;
PROC SQL expects the view to reside in the same
SAS library as the contributing table or tables.

S:\workshop\data

sq
48
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Scenario
s:\workshop\data s:\workshop

sq.customer totalcustomer

sq mkt

Members of the marketing team

moved the view from the sq library
to their working location.

49
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-35

Scenario

libname mkt "s:/workshop";

proc sql; marketing

select *
from mkt.totalcustomer
quit; Marketing executes
a query using the
ERROR: File MKT.CUSTOMER.DATA does not exist. view but receives
this error.
Why?

50
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Exploring the Problem

proc sql;
create view sq.totalcustomer as
select State,count(*) as TotalCustomer
from customer
group by State;
quit;

s:\workshop\data s:\workshop

customer customer

The view assumes that the customer

data is in the s:\workshop library.
sq mkt
51
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

This violated the one-level naming convention in the CREATE VIEW query. When marketing used
the view, it retrieved the stored query. The stored query that we are using uses the one-level naming
convention. When the view was created, PROC SQL assumed that the customer data was in the sq
library. When marketing moved the view and executed a query using the view, the stored query
assumed that the customer table is now in the mkt library in s:\workshop. However, the customer
table is not in that location.

Making a View Portable

CREATE VIEW view-name AS SELECT …

USING LIBNAME-clause<, …LIBNAME-clause>;
libname mkt "s:/workshop";
proc sql;
create view mkt.totalcustomer as
select State,count(*) as TotalCustomer
from sq.customer
group by State
using libname sq 's:/workshop/data';
quit;

The scope of the libref is local to the view and does not conflict
with any identically named librefs in the SAS session.
52
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

By using a SAS enhancement, you can create a usable view that is stored in a different physical
location than its source tables. In other words, you can make the view portable.
You can embed a SAS LIBNAME statement or a SAS/ACCESS LIBNAME statement in a view by
using the USING LIBNAME clause. When PROC SQL executes the view, the stored query assigns
the libref. For SAS/ACCESS librefs, PROC SQL establishes a connection to a DBMS. The scope of
the libref is local to the view and does not conflict with any identically named librefs in the SAS
session. When the query finishes, the libref is disassociated. The connection to the DBMS is
terminated, and all data in the library becomes unavailable.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-37

Views

Advantages
• avoid storing copies of large tables
• avoid a frequent refresh of table copies; when the underlying data
changes, a view surfaces the most current data
• combine data from multiple database tables and multiple libraries or
databases
• simplify complex queries
• prevent other users from inadvertently altering the query code

53
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Views

Disadvantages
• Views might produce different results each time they are accessed if the
data in the underlying data sources changes.
• Views can require significant resources each time that they execute.
With a view, you save disk storage space at the cost of extra CPU and
memory usage.

54
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Syntax Summary

SELECT …
FROM (SELECT col-name
FROM …
<WHERE …>) ;
In-Line View
CREATE VIEW table-name AS query
CREATE VIEW
CREATE VIEW …
USING LIBNAME libref engine "path";
USING Clause
55
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-39

Practice

Level 1
6. In-Line View Summarizing CreditScore
Determine which customers have an extremely high credit score relative to other customers in
their ZIP code (Zip). Similar to the practice in the previous section, extremely high credit is
defined as greater than 2 standard deviations above the mean of CreditScore. However, rather
than use an overall high-credit threshold for all customers, we need to calculate the threshold for
each value of Zip. This can be accomplished using an in-line view.
a. Open s104p06.sas from the practices folder. Run the query to summarize the
HighZipCredit threshold for each Zip value for the first 1000 rows.
Partial Results

b. Use the query from step a as an in-line view to join with the sq.customer table.
1) Select c.CustomerID, c.Zip, and c.CreditScore from the sq.customer table, and select
s.HighZipCredit from the in-line view. Format the c.Zip column using the Z5. format.
2) Perform an inner join with the sq.customer table and the in-line view from step a. Give
the sq.customer table the alias c and the in-line view the alias s. Remove the INOBS=
option from the in-line view.
3) Use c.Zip = s.Zip as the join criteria.
4) Filter rows where the customer's c.CreditScore value is greater than the
s.HighZipCredit value.
5) Order the results by Zip and descending CreditScore.
6) Add an appropriate title to the report.

Partial Results

c. What is the last Zip value and the corresponding CreditScore value in your final report?

Level 2
7. Building a Complex Query with In-Line Views
Determine which employees have the highest salary for their job title in every state. Using the
sq.employee table, write a query to calculate the maximum Salary value for each value of
JobTitle within each state.
a. Using the sq.employee table, write a query to calculate the maximum salary for each job title
within each state.
1) Convert the values of the State column to uppercase and name the column State to
standardize the state code values. Select JobTitle and calculate the maximum salary.
Name the column MaxJobSalary.
2) Filter rows where the State is not null.
3) Group the results by State and JobTitle.
4) Order the results by State.
5) Add an appropriate title.
6) Run the query and compare your results.
Partial Results

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.2 In-Line View s (Query in the FROM Clause) 4-41

b. Using the query in step a as an in-line view, create a report by joining that result with the
sq.employee table to do the following:
1) Display EmployeeID, EmployeeName, State, JobTitle, and Salary for the highest paid
employee for each value of JobTitle in every state.
2) Perform an inner join with the sq.employee table and the in-line view from step a. Give
the sq.employee table the alias detail and the in-line view the alias summary.
Hint: Remove the ORDER BY clause when using an in-line view.
3) Use JobTitle, State, and Salary equal to MaxJobSalary as the join criteria.
4) Order the report by State and JobTitle.
5) Add an appropriate title and format the Salary values with a dollar sign and comma.
Partial Results

c. Who is the last employee in the final report?

Challenge
8. Building a Complex Query Using a Join and Subquery
Generate a report of the total estimated number of individuals next year who made or received
digital payments in the past year (% age 15+) in South Asia to determine the best country to
promote a digital payment app.
a. Calculate the total estimated population for ages 15+ for countries in South Asia in the
sq.globalpop table.
1) Select the CountryCode, sum of EstYear1. Name the new column EstYear1Pop and
format using commas.
2) Filter the SeriesName column for estimated population greater than 15 years of age.
Use a subquery from the sq.globalmetadata table to include only CountryCode in
South Asia.
Hint: Use the LIKE operator in the WHERE clause to include rows for ages 15+.
3) Group by CountryCode.
4) Run the query and compare your results.

Results

b. Use the previous query as an in-line view to join by CountryCode and multiply the estimated
percentage of individuals who made or received digital payments in the past year (% age
15+) by the estimated population for next year.
1) Select the CountryCode and IndicatorName columns from the sq.globalfindex table.
Convert the EstYear1 value from sq.globalfindex to a percentage by dividing by 100
and then multiply by the EstYear1Pop value from the in-line view. Name the column
Estimate and format using commas.
Note: Multiplying the EstYear1 percentage of individuals by the EstYear1Pop total
population for the country returns an estimated number of individuals who used
digital payments.
2) Perform an inner join of sq.globalfindex and the in-line view from step a. Assign
sq.globalfindex the alias f, and assign the in-line view the alias pop. Use CountryCode
as the join criteria.
3) Filter the rows by the IndicatorName value of Made or received digital payments in the
past year (% age 15+).
4) Order the results by Estimate descending. Add an appropriate title.
Results

c. Which value of CountryCode has the highest estimated use of digital payments?

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-43

4.3 Subquery in the SELECT Clause

Subquery in the SELECT Clause

Outer Query
SELECT col-name, (SELECT col-name
FROM … A subquery in the
<WHERE …>) SELECT clause can
FROM … return a single value
to the outer query.
<WHERE …>
<GROUP BY …>
<HAVING …>
<ORDER BY …>;

58
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Scenario
12.0%

8.56% 6.02% State / Total

PopEstimate1 sum(PopEstimate1)

Create results that show

next year's estimated
percentage of population
of each state.

59
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery in the SELECT Clause

select Name, PopEstimate1/(select sum(PopEstimate1)

from sq.statepopulation)
as PctPop format=percent7.2
from sq.statepopulation
order by PctPop desc;

State
PopEstimate1
/ 326477837

PopEstimate1 from total estimated

each state population for the US
60
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

When using a subquery in the SELECT clause, you do not have to use a value from the table in the
outer query. You can retrieve values from other tables.

Remerging Summary Statistics

selects each name and
summarizes the entire
PopEstimate1 value from
column
each row

The remerge
select Name, PopEstimate1 / sum(PopEstimate1)
feature of SAS
as PctPop format=percent7.2
makes two from sq.statepopulation
passes through a order by PctPop desc;
table.

61
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-45

Aggregate functions, such as the SUM function, can cause the same calculation to repeat for every
row. This occurs whenever PROC SQL remerges data. Remerging occurs whenever any of the
following conditions exist:
• The SELECT clause references a column that contains an aggregate function and other columns
that are not listed in the GROUP BY clause.
• The ORDER BY clause references a column that is not referenced by the SELECT clause.
When a query remerges data, PROC SQL displays a note in the log to indicate that data remerging
has occurred. PROC SQL runs an internal query to find the sum and then runs another internal
query to divide each state's population by the sum.

Remerging Summary Statistics

Scenario
Remerge summary statistics in SAS to find the percentage population of each state.

Files
• s104d04.sas
• statepopulation – a SAS table that contains estimated state populations for the next three years

Syntax

PROC SQL;
SELECT col-name, summary function(column)
FROM table;
QUIT;

Notes
• The SELECT clause references a column that contains an aggregate function and other columns
that are not listed in the GROUP BY clause.

Demo
1. Open the s104d04.sas program in the demos folder and find the Demo section. Run the query
to select the Name and PopEstimate1 columns from the sq.statepopulation table.
proc sql;
select Name, PopEstimate1
from sq.statepopulation;
quit;
2. In the SELECT clause, add the SUM function to sum PopEstimate1. Format the column using
the COMMA12 format. Run the query and examine log and results.
Note: All values are the sum of the entire PopEstimate1 column, whereas the Name and
PopEstimate1 columns have individual rows from the input table.
proc sql;
select Name, PopEstimate1, sum(PopEstimate1) format=comma12.
from sq.statepopulation;
quit;

3. Modify the SELECT clause by dividing PopEstimate1 and sum(PopEstimate1). Replace the
COMMA12. format with the PERCENT7.2 format. Name the new column PctPop. Run the query
and examine the log and results.
proc sql;
select Name,
PopEstimate1/sum(PopEstimate1) as PctPop format=percent7.2

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-47

from sq.statepopulation;
quit;
4. Add an ORDER BY clause and sort the results by descending PctPop. Run the query and
examine the results.
proc sql;
select Name,
PopEstimate1/sum(PopEstimate1) as PctPop format=percent7.2
from sq.statepopulation
order by PctPop desc;
quit;

Remerging Summary Statistics

select Region,
sum(PopEstimate1) as TotalRegion format=comma14.
from sq.statepopulation;

Specifying a column and a summary function

remerges the data and does not return the
desired result.

63
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Remerging can be a powerful tool. However, it does not always produce the desired result. The most
common example is when you forget the GROUP BY clause in a query and specify a grouping
column and a summary function. Always check your log to help avoid this error and others.

Disabling the Remerging of Summary Statistics

PROC SQL NOREMERGE;

proc sql noremerge;

select Region,
sum(PopEstimate1) as TotalRegion format=comma14.
from sq.statepopulation;
quit;

ERROR: The query requires remerging summary statistics back with the
original data. This is disallowed due to the NOREMERGE proc option
or NOSQLREMERGE system option.

64
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Resubmitting the query with the NOREMERGE option in the PROC SQL statement produces no
output and results in an error message in the SAS log.

4.05 Activity
Open s104a05.sas from the activities fol der a nd perform the following tasks
to di s a ble remerging of s ummary s tatistics:
1. Exa mi ne a nd run the query. Exa mi ne the l og a nd the results. What note
do you s ee i n the l og?
2. Add the PROC SQL option NOREMERGE. Run the query. Di d i t run
s ucces sfully? Wha t wa s the error i n the l og?
3. Add a GROUP BY cl a use after the FROM cl a use a nd group by Region. Run
the query. Di d i t run s ucces sfully?

65
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-49

Scenario

How can we calculate the

percentage of population in
each region?

68
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Remerging GROUP BY Summary Statistics

proc sql;
select Name, PopEstimate1,
sum(PopEstimate1) as TotalRegionEst format=comma14.,
PopEstimate1/calculated TotalRegionEst as
PctRegion format=percent7.1,
Region SAS remerges the sum
from sq.statepopulation
group by Region of each region.
order by Region, PctRegion desc;
quit;

69
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Syntax Summary

SELECT col-name, (SELECT col-name

FROM … PROC SQL NOREMERGE;
<WHERE …>) PROC SQL Option
FROM table1;
Subquery in the SELECT Clause

SELECT col-name, summary function(column)

FROM table1;
Remerging Summary Statistics

70
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Beyond SQL Essentials

What if you want to ...

. . . l ea rn a bout using a . . . l ea rn a bout . . . revi ew exa mples

refl exive join wi th a bui l ding s ubqueries i n us i ng s ubqueries to
s ubquery? SAS Enterpri s e Guide? s el ect da ta?

• View the SAS paper Nifty • Read the SAS blog Building • Visit Using Subqueries to
Uses of SQL Reflexive Join an SQL subquery in SAS Select Data in the SAS
and Sub-query in SAS. Enterprise Guide. documentation.

71
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-51

Practice

Level 1
9. Remerging Summary Statistics
Determine which states have the most estimated births for next year. Include a column showing
each state's births as the percent of national births.
a. Select the Name and Births1 columns. Create a new column named PctBirth by dividing
Births1 for each state by the sum of Births1 for all states. Format the new column using the
PERCENT format.
b. Order the results by descending PctBirth.
c. Add an appropriate title.
Partial Results

d. Which state has the highest percentage of estimated births for next year?

Level 2
10. Subquery in the SELECT Clause with an In-Line View
Find the top 10 countries that have the highest percentage of estimated global population using
the population estimates of ages 15+ in the sq.globalfull table.
a. Use the sq.globalfull table to write a query to sum the EstYear1Pop of all countries.
1) Use an in-line view to select the distinct CountryCode and EstYear1Pop values from
the sq.globalfull table.
Note: You must find the distinct estimated population of each country because the
data contains the estimated population for each country multiple times.
2) Sum the EstYear1Pop column and name the column new column EstPct. Format the
column using the COMMA format.
3) Run the query and compare your results.

Results

b. Create a new query using the query from step a in the SELECT clause to determine the
estimated global population percentage of each country.
1) Select the distinct CountryCode and ShortName values. Determine the percentage
population by creating a new column. Divide the EstYear1Pop value of each country by
the total value calculated in step a. Name the new column PctPop and format using the
PERCENT format.
2) Use the sq.globalfull table.
3) Order the results by PctPop descending.
4) Limit the results to the top 10 countries.
5) Add an appropriate title.
Results

c. Which country has the highest estimated population of individuals 15 or over?

Challenge
11. Remerging GROUP BY Summary Statistics
In the sq.statepopulation table, states in the United States are categorized by divisions.
Calculate the percentage of births for next year by each state for its division.
a. Use the sq.statepopulation table to write a query to sum the values of Births1 for each
division and divide the summarized value by the Births1 values of each state.

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4.3 Subquery in the SELECT Clause 4-53

1) Select the Name, Division, and Births1 columns. Create two new calculated columns.
a) The first column should sum all Births1 values. Name the column TotalDivisionEst,
and format it using the COMMA format.
b) The second column should divide Births1 by the new column, TotalDivisionEst.
Name the column PctDivision, and format it using the PERCENT format.
2) Group the query by Division.
3) Order the results by Division and PctDivision descending.
4) Format the Births1 column using the COMMA format.
5) Add an appropriate title.
Partial Results

b. Which state has the highest percentage of estimated births in Division 8?

4.4 Solutions
Solutions to Practices
1. Subquery That Returns a Single Value
/*s104s01.sas*/

/*a*/
proc sql;
select mean(nPopChg3)
from sq.statepopulation;
quit;

/*b*/
title "States with an estimated 3-Year Population Growth";
title2 "Greater than the Overall Average";
proc sql;
select Name, nPopChg3 label="Estimated Growth" format=comma16.
from sq.statepopulation
where nPopChg3 > (select mean(nPopChg3)
from sq.statepopulation)
order by nPopChg3 desc;
quit;
title;
Which state has the lowest estimated growth in your results? MA, with an estimated growth of
38,903
2. Subquery with Multiple Functions
/*s104s02.sas*/

/*a*/
proc sql;
select sum(avg(CreditScore),(2*std(CreditScore))) as HighScore
from sq.customer
where state='TX';
quit;

/*b*/
title 'Texas Customers with Higher than Average Credit Scores';
proc sql;
select CustomerID, FirstName, LastName, CreditScore
from sq.customer
where state="TX" and
CreditScore >
(select sum(avg(CreditScore),(2*std(CreditScore)))
as HighScore
from sq.customer
where state='TX')

order by CreditScore desc;

quit;
title;
In your report, what is the lowest value for CreditScore? 766
3. Subquery Using Different Tables
/*s104s03.sas*/

/*a*/
proc sql;
select CountryCode
from sq.globalmetadata
where upcase(Region)='EUROPE & CENTRAL ASIA' and
upcase(IncomeGroup)='HIGH INCOME';
quit;

/*b*/
title 'Estimated Forecasted Percentages';
title2 'Countries in Europe & Central Asia with High Incomes';
title3 'Borrowed Money for Health or Medical Purposes';
proc sql;
select CountryCode,
EstYear1/100 as EstPct1 format=percent7.2,
EstYear3/100 as EstPct3 format=percent7.2
from sq.globalfindex
where IndicatorName='Borrowed for health or medical purposes
(% age 15+)' and
CountryCode in
(select CountryCode
from sq.globalmetadata
where upcase(Region)='EUROPE & CENTRAL ASIA'
and upcase(IncomeGroup)='HIGH INCOME')
order by EstYear1 desc;
quit;
title;
Which country in your report has the lowest EstPct1 value for next year? FIN
4. Using a Subquery with Summarized Data
/*s104s04.sas*/

title "Customers with Lower than Average Utility Payments";

proc sql;
select CustomerID, avg(Amount) as UtilityAmt
from sq.transactionfull
where Type="Utilities"
group by 1
having UtilityAmt <
(select avg(Amount)
from sq.transactionfull

where Type="Utilities")
order by UtilityAmt desc;
quit;
title;
How many customers in the sq.transaction table have lower than average utility payments? Five
5. Using Nested Subqueries
title "Countries with a Higher Population Estimate than the Mean";
title2 "Number of Outstanding Housing Loan Increases from Year1 to
Year3";

/*b*/
proc sql;
select CountryCode
from CountryEstPop
where EstPop > (select mean(EstPop)
from CountryEstPop);
quit;

/*c*/
proc sql;
select CountryCode, IndicatorName,
EstYear1/100 as EstYear1 format=percent7.2,
EstYear3/100 as EstYear3 format=percent7.2,
calculated EstYear3 - calculated EstYear1
as PctIncrease format=percent7.2
from sq.globalfindex
where IndicatorName = "Outstanding housing loan (% age 15+)"
and EstYear1 is not null
and EstYear3 is not null
and EstYear1 < EstYear3
and CountryCode in
(select CountryCode
from CountryEstPop
where EstPop > (select mean(EstPop)
from CountryEstPop))
order by PctIncrease desc;
quit;
title;
Which country had the largest value for PctIncrease? BGD

6. In-Line View Summarizing CreditScore

/*s104s06.sas*/

title 'Customers with Extremely High Credit for their Zipcode';

proc sql;
select c.CustomerID, c.Zip format=z5., c.CreditScore,
s.HighZipCredit
from sq.customer as c inner join
(select Zip,
sum(avg(CreditScore),(2*std(CreditScore)))
as HighZipCredit
from sq.customer
where CreditScore is not null
group by Zip) as s
on c.Zip=s.Zip
where c.CreditScore>s.HighZipCredit
order by Zip, CreditScore desc;
quit;
title;
What is the last Zip value and the corresponding CreditScore value in your final report?
The last Zip value is 99701 with a CreditScore value of 815.
7. Building a Complex Query with In-Line Views
/*s104s06.sas*/

/*a*/
title 'Maximum Salary for Each Job in every State';
proc sql;
select upcase(State) as State, JobTitle,
max(Salary) as MaxJobSalary
from sq.employee
where State is not null
group by State, JobTitle
order by State;
quit;
title;

/*b*/
title 'Employees with Highest Salary for their Job in every State';
proc sql;
select detail.EmployeeID, detail.EmployeeName, detail.State,
detail.JobTitle, detail.Salary format=dollar12.
from sq.employee as detail inner join
(select upcase(State) as State,
JobTitle, max(Salary) as MaxJobSalary
from sq.employee
where State is not null
group by State, JobTitle) as summary
on detail.Jobtitle=summary.JobTitle and

detail.State=Summary.State and
detail.Salary=Summary.MaxJobSalary
order by detail.State, detail.JobTitle;
quit;
title;
Who is the last employee in the final report? The last employee is Pongor, Katherine .
8. Building a Complex Query Using a Join and Subquery
/*s104s07.sas*/

/*a*/
title 'Estimated Population for Next Year in South Asia ';
title2 'Ages 15+';
proc sql;
select CountryCode,
sum(EstYear1) as EstYear1Pop format=comma14.
from sq.globalpop
where SeriesName not like '%00-04' and
SeriesName not like '%05-09' and
SeriesName not like '%10-14' and
CountryCode in (select CountryCode
from sq.globalmetadata
where Region = 'South Asia')
group by CountryCode;
quit;
title;

/*b*/
title 'Estimated Population who Made or Received Digital Payments
in the Past Year';
title2 'South Asia Ages 15+';
proc sql;
select f.CountryCode, f.IndicatorName,
((f.EstYear1/100) * pop.EstYear1Pop)
as Estimate format=comma14.
from sq.globalfindex as f inner join
(select CountryCode, sum(EstYear1)
as EstYear1Pop format=comma14.
from sq.globalpop
where SeriesName not like '%00-04' and
SeriesName not like '%05-09' and
SeriesName not like '%10-14' and
CountryCode in (select CountryCode
from sq.globalmetadata
where Region = 'South Asia')
group by CountryCode) as pop
on f.CountryCode = pop.CountryCode
where IndicatorName = 'Made or received digital payments in
the past year (% age 15+)'

order by Estimate desc;

quit;
Which value of CountryCode has the highest estimated use of digital payments?
IND has an estimated digital payment use of 177,123,495 people.
9. Remerging Summary Statistics
title 'Estimate Percentage of Births by Each State';
proc sql;
select Name, Births1,
Births1/sum(Births1) as PctBirth format=percent7.2
from sq.statepopulation
order by PctBirth desc;
quit;
title;

/*Alternate Solution*/
title 'Estimate Percentage of Births by Each State';
proc sql;
select Name, Births1,
Births1/(select sum(Births1)
from sq.statepopulation)
as PctBirth format=percent7.2
from sq.statepopulation
order by PctBirth desc;
quit;
title;
Which state has the highest percentage of estimated births for next year? CA (California)
10. Subquery in the SELECT Clause with an In-Line View
/*s104s10.sas*/

/*a*/
title 'Estimated Population of Ages 15+ for Next Year';
proc sql;
select sum(EstYear1Pop) as EstPct format=comma16.
from (select distinct CountryCode, EstYear1Pop
from sq.globalfull);
quit;
title;

/*b*/
title 'Top 10 Countries by Estimate Population';
title2 'Ages 15+';
proc sql outobs=10;
select distinct CountryCode, ShortName,
EstYear1Pop/
(select sum(EstYear1Pop) as EstPct format=comma16.
from (select distinct CountryCode, EstYear1Pop
from sq.globalfull))

as PctPop format=percent7.2
from sq.globalfull
order by PctPop desc;
quit;
title;
Which country has the highest estimated population of individuals 15 or over? China
11. Remerging GROUP BY Summary Statistics
/*s104s11.sas*/

title "Percentage of Estimated Births by Each Division";

proc sql;
select Name, Division, Births1 format=comma14 .,
sum(Births1) as TotalDivisionEst format=comma14.,
Births1/calculated TotalDivisionEst
as PctDivision format=percent7.1
from sq.statepopulation
group by Division
order by Division, PctDivision desc;
quit;
title;
Which state has the highest percentage of estimated births in Division 8?
AZ with 27.7% of estimated births

Solutions to Activities and Questions

continued...
4.01 Activity – Correct Answer
2.

…
where PopEstimate1 > (select avg(PopEstimate1),
"Average Estimated Population"
from sq.statepopulation);

3. Run the query. Wha t i s the s yntax error i n the l og?

ERROR: A subquery cannot select more than one column.

ERROR: A Composite expression (usually a subquery) is used incorrectly in an
expression.

12
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.01 Activity – Correct Answer

…
where PopEstimate1 > (select avg(PopEstimate1),
'Average Population'
from sq.statepopulation);

A subquery must return

values in a single column.

13
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.02 Activity – Correct Answer

4. How ma ny di vi s ions have a hi gher a verage PopEstimate1 tha n the
a vera ge PopEstimate1 of a l l the s tates? Five divisions

select Division, avg(PopEstimate1) as avgDivisionPop

from sq.statepopulation
group by Division
having avgDivisionPop > (select avg(PopEstimate1)
from sq.statepopulation);

Subquery

17
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.03 Activity – Correct Answer

select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 < any (select PopEstimate1
from sq.statepopulation
where Name in ("NY","FL"));

select Name, PopEstimate1

from sq.statepopulation
where PopEstimate1 < (select max(PopEstimate1)
from sq.statepopulation
where Name in ("NY","FL"));

49 states
26
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.04 Activity – Correct Answer

1. Crea te a vi ew na med VWtotalcustomer from the query. Run the query
a nd exa mine the log.

create view VWtotalcustomer as

select State,count(*) as TotalCustomer
from sq.customer
group by State;

2. Run the code i n the s ecti on Use the View in the PROCS Below. Whi ch
s tate has the hi ghest number of cus tomers? California

47
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

continued...
4.05 Activity – Correct Answer
1. Wha t note do you s ee in the log?

NOTE: The query requires remerging summary statistics back with the
original data.

2. Di d i t run s uccessfully? Wha t wa s the error i n the l og? No. The

NOREMERGE option disables remerging of summary statistics.

ERROR: The query requires remerging summary statistics back with the original
data. This is disallowed due to the NOREMERGE proc option or NOSQLREMERGE
system option.

66
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

4.05 Activity – Correct Answer

3. Add a GROUP BY cl a use after the FROM cl a use a nd group by Region. Run
the query. Di d i t run s ucces sfully? Yes. The data is not remerged when
you use the GROUP BY clause.

proc sql noremerge;

select Region,
sum(PopEstimate1) as TotalRegion format=comma14.
from sq.statepopulation
group by region;
quit;

67
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
Lesson 5 Set Operators
5.1 Introduction to Set Operators ...................................................................................... 5-3

5.2 INTERS ECT, EXCEPT, and UNION ................................................................................ 5-9

Demonstration: Using the UNION Set Operator......................................................... 5-18
Practice............................................................................................................... 5-24

5.3 OUTER UNION ........................................................................................................... 5-28

Demonstration: Using the OUTER UNION Set Operator............................................. 5-30
Practice............................................................................................................... 5-34

5.4 Solutions ................................................................................................................... 5-36

Solutions to Practices ............................................................................................ 5-36
Solutions to Activities and Questions........................................................................ 5-39
5-2 Lesson 5 Set Operators

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
5.1 Introduction to Set Operators 5-3

5.1 Introduction to Set Operators

Combining Data Using Joins

Inner Join Returns only
rows that match

Matches and Matches and all Matches and

nonmatches nonmatches nonmatches
Left Join Full Join Right Join

3
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subsetting Data Using Subqueries

Outer Query
SELECT col-name, col-name
FROM input-table
WHERE column operator (SELECT col-name
FROM input-table…);
Subquery
Combine data
(select avg(PopEstimate1)
using multiple, from sq.statepopulation
nested queries, where … )…
called subqueries.

4
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Combining Data Using Set Operators

Intermediate

Res ult
Query 1
Set 1

Set Operators

Res ult
Query 1
Set 2

5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

A set operator vertically combines the intermediate result sets from two queries to produce a final
result set.

Business Data

Target Customer List: Email Respondents: Phone Respondents:

saleslist salesemail salesphone

List of targeted customers for a new promotional credit card.

Customers are employed, over x dollars….
6
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
5.1 Introduction to Set Operators 5-5

Scenario

Which customers Which customers

have responded to have not responded
both email and to the email
phone requests? request?

Which customers
responded to Complete list of all
either phone or customer responses
email?

Target Customer List Table

saleslist

List of target
Target Customer List customers and their
contact information.

Response Tables

Customer responses
by email and phone.
Email Respondents Phone Respondents

salesemail salesphone

Scenario
saleslist salesemail

salesphone

All tables contain CustomerID.

The CustomerID column exists in all tables.

Default Behavior of Rows in Set Operators

all rows in the
unique rows unique rows in query results
common in both the first query all unique rows
queries only
Outer Union

Intersect Except Union

Default Behavior of Columns in Set Operators

Columns aligned by Includes all columns

position from both result sets

Outer Union
Intersect Except Union

Using Set Operators

Query result set 1

You must place a

semicolon after the
SELECT query…
last SELECT
UNION | EXCEPT | INTERSECT | OUTER UNION
statement.
SELECT query…;

Query result set 2

You must place a semicolon after the last SELECT statement only.

Using Set Operators

ALL modifies the default

behavior for rows.

SELECT query…
UNION | EXCEPT | INTERSECT | OUTER UNION <ALL> <CORR>
SELECT query…;

CORR modifies the default

behavior for columns.

5.2 INTERSECT, EXCEPT, and UNION

INTERSECT Operator
salesemail salesphone
CustomerID EmailResp CustomerID SalesRep PhoneResp
1939774314 Accepted 1939774314 121038 Declined
1999302252 Declined 1999302252 120145 Call Back Which customers
1963960449 Declined 1999302252 120145 Accepted have responded
1908694347 Accepted 1963960449 120145 Declined
to both email and
phone sales?
1960311448 Accepted 1987175132 120145 Declined
1905044343 Declined 1970095442 121137 Accepted

INTERSECT Operator

SELECT query… CustomerID

INTERSECT <ALL> <CORR> 1939774314
SELECT query…; 1999302252
Result Set 1 1963960449
proc sql; …
select CustomerID
from sq.salesemail
intersect
select CustomerID CustomerID
from sq.salesphone;
quit; 1939774314
1999302252
Result Set 2
1999302252
…
17
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Default Behavior of the INTERSECT Operator

Result Set 1 Result Set 2
CustomerID CustomerID
1939774314 1939774314
1999302252 1999302252
1963960449 1999302252
1908694347 1963960449
1960311448 1987175132
1905044343 1970095442

First, duplicate rows are removed

from each of the intermediate result
sets if they exist.

Default Behavior of the INTERSECT Operator

Result Set 1 Result Set 2
CustomerID CustomerID
1939774314 1939774314
1999302252 1999302252
1963960449 1999302252
1908694347 1963960449
1960311448 1987175132
1905044343 1970095442

Rows from the first intermediate

result set that are also in the second
intermediate result set are selected.

INTERSECT Operator
title "Customers Who Have Responded to
Both Email and Phone Sales";

proc sql;
select CustomerID
from sq.salesemail
intersect
select CustomerID
from sq.salesphone;
quit;

5.01 Activity
Open s105a01.sas from the activities fol der a nd perform the following to
fi nd uni que cus tomers who ha ve responded by phone a nd email:
1. Run the fi rs t queries to previ ew the sq.salesemail a nd sq.salesphone
tabl es. Exa mine the col umns i n both tables.
2. In the Intersect s ecti on, exa mine a nd run the query. Di d the query run
s ucces sfully? Why not?
3. Add the CORR keyword a fter the INTERSECT s et operator. Run the query.
Di d the query run s uccessfully? Why?

Set Operators and Joins

proc sql; proc sql;

select CustomerID select distinct e.CustomerID
from sq.salesemail from sq.salesemail as e inner join
intersect sq.salesphone as p
select CustomerID on e.CustomerID = p.CustomerID;
from sq.salesphone; quit;
quit;
You can use an inner
join to produce
identical results.

You can use an INNER JOIN to produce identical results. In this example, we use an inner join to
find all matches of the two tables. The DISTINCT keyword removes any duplicate rows resulting in
the same result as the INTERSECT set operator.

EXCEPT Operator
saleslist salesemail
UserID CustomerID … CustomerID EmailResp
[email protected] 1939774314 … 1939774314 Accepted
[email protected] 1958716829 … 1999302252 Declined
[email protected] 1999302252 … 1963960449 Declined
[email protected] 1963960449 … 1908694347 Accepted
List of customers
1960311448 Accepted
who have not
[email protected] 1987175132 … 1905044343 Declinedresponded to our
[email protected] 1970095442 …
sales email
[email protected] 1908694347 …
barmablanton521@n/a.com 1918638906 …
[email protected] 1960311448 …
[email protected] 1905044343 …
25
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

EXCEPT Operator

SELECT query… CustomerID

EXCEPT <ALL> <CORR> 1939774314
SELECT query…; 1958716829
Result Set 1 1999302252
proc sql;
…
select CustomerID
from sq.saleslist
except
select CustomerID
from sq.salesemail; CustomerID
quit; 1939774314
Result Set 2 1999302252
1963960449

Default Behavior of the EXCEPT Operator

Result Set 1 Result Set 2
CustomerID CustomerID
1939774314 1939774314
1958716829 1999302252
1999302252 1963960449
1963960449 1908694347
1987175132 1960311448
1970095442 1905044343
1908694347
1918638906 First, duplicate rows are removed from
1960311448 each of the intermediate result sets if they
1905044343 exist.

Default Behavior of the EXCEPT Operator

Result Set 1 Result Set 2
CustomerID CustomerID
1939774314 1939774314
1958716829 1999302252
1999302252 1963960449
1963960449 1908694347
1987175132 1960311448
1970095442 1905044343
1908694347
1918638906 Rows from the first intermediate result set
1960311448 that are not in the second intermediate
1905044343 result set are selected.

EXCEPT Operator
title "Customers Who Haven't Responded to the Sales Email";
proc sql;
select CustomerID
from sq.saleslist
except
select CustomerID
from sq.salesemail;
quit;

5.02 Activity
Open s105a02.sas from the activities fol der a nd perform the following to
fi nd a ll target cus tomers who have not responded to our s a les phone ca ll:
1. Run the fi rs t queries to previ ew the sq.saleslist a nd sq.salesphone
tabl es. Exa mine the col umns i n both tables.
2. Compl ete the query to fi nd a ll customers from the sq.saleslist tabl e who
ha ve not res ponded to our s ales ca ll i n sq.salesphone.
3. How ma ny cus tomers ha ve not res ponded to our phone ca ll?

Set Operators and Subqueries

select distinct CustomerID

from sq.saleslist
where CustomerID not in(select CustomerID
from sq.salesphone);

select CustomerID You can use a

from sq.saleslist subquery to produce
except
select CustomerID identical results.
from sq.salesphone;

You can use a subquery to produce identical results. In this example, we use a subquery to return all
CustomerID values in the salesphone table. Then we use the WHERE clause to subset for all
CustomerID values in the saleslist table that are not in the list returned by our subquery. The
DISTINCT keyword removes any duplicate rows, resulting in the same result as the EXCEPT set
operator.

UNION Operator
salesemail salesphone
CustomerID EmailResp CustomerID SalesRep PhoneResp
1939774314 Accepted 1939774314 121038 Declined
1999302252 Declined 1999302252 120145 Call Back
1963960449 Declined 1999302252 120145 Accepted Find the total
number of unique
1908694347 Accepted 1963960449 120145 Declined
customers who
1960311448 Accepted 1987175132 120145 Declined
responded to either
1905044343 Declined 1970095442 121137 Accepted phone or email.
1955264298 121038 Declined

UNION Operator

SELECT query… CustomerID

UNION <ALL> <CORR> 1939774314
SELECT query…; 1999302252
Result set 1 1963960449
proc sql;
…
select CustomerID
from sq.salesemail
union
select CustomerID CustomerID
from sq.salesphone; 1939774314
quit;
1999302252
Result set 2 1999302252
…
34
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Default Behavior of the UNION Operator

CustomerID
1939774314
1999302252
Result set 1 1963960449
1908694347
1960311448
1905044343
Combines result sets
1939774314
1999302252
Result set 2 1999302252
1963960449
1987175132
1970095442
35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Default Behavior of the UNION Operator

CustomerID
1939774314
1999302252 CustomerID
Result set 1 1963960449 1905044343
1908694347 1908694347
1960311448 1939774314
1905044343 Removes duplicate values 1960311448
and returns unique rows 1963960449
1939774314
1970095442
1999302252
1987175132
Result set 2 1999302252
1999302252
1963960449
1987175132
1970095442
36
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Using the UNION Set Operator

Scenario
Use the UNION set operator to count the number of unique customers who responded to an email or
a phone sales attempt.

Files
• s105d01.sas
• salesemail – a SAS table that contains email response information from targeted customers
• salesphone – a SAS table that contains phone response information from targeted customers

Syntax

PROC SQL;
SELECT query…
UNION <ALL> <CORR>
SELECT query…;
QUIT;

Notes
• The UNION set operator produces all unique rows from both queries.
• The CORR keyword overlays columns that have the same name in both tables. When used with
EXCEPT, INTERSECT, and UNION, CORR suppresses columns that are not in both tables.

Demo
1. Open the s105d01.sas program in the demos folder and find the Demo section. Under Explore
the SALESEMAIL and SALESPHONE Tables, run the two queries. Discuss the tables and the
desired outcome.
proc sql;
select *
from sq.salesemail;
select *
from sq.salesphone;
quit;
2. In the next section, complete the query to find all unique customers who responded to either an
email or phone call. Begin with the SELECT statement and select all columns from the
sq.salesemail table.
proc sql;
select *
from sq.salesemail
quit;

3. Use the UNION set operator followed by another SELECT statement to select all columns from
the sq.salesphone table. Run the query and examine the syntax error.
proc sql;
select *
from sq.salesemail
union
select *
from sq.salesphone;
quit;
4. Add the keyword CORR after the UNION set operator. Run the query and examine the log and
results.
Note: The CORR keyword aligns the columns that have the same name in both tables and
removes any columns not found in both tables.
proc sql;
select *
from sq.salesemail
union corr
select *
from sq.salesphone;
quit;
5. Remove the CORR keyword, and specify the CustomerID column in both SELECT clauses. Run
the query and examine the log and results.
proc sql;
select CustomerID
from sq.salesemail
union corr
select CustomerID
from sq.salesphone;
quit;
6. Add another SELECT statement at the first line in the SQL procedure. Use the COUNT(*)
function to count all rows. Name the column TotalNum. Add a FROM clause and use the
previous query as a subquery in the FROM clause (in-line view). Be sure to add parentheses
around the subquery. Run the query and examine the results.
proc sql;
select count(*) as TotalNum
from (select CustomerID
from sq.salesemail
union
select CustomerID
from sq.salesphone);
quit;

Default Behavior of the UNION Operator

UNION INTERSECT/EXCEPT

Combi nes Removes

res ul t s ets dupl ica te rows

Removes Combi nes

dupl ica te rows res ul t s ets

The UNION set operator works in a different order than the INTERSECT and EXCEPT operators.
The UNION set operator first combines results sets and then removes duplicate rows. INTERSET
and EXCEPT remove duplicate rows and then combine result sets.

Default Behavior of the UNION Operator

select CustomerID, EmailResp
from sq.salesemail
union
select CustomerID,PhoneResp, SalesRep
from sq.salesphone;

If the result sets have different numbers of columns,

SAS extends one table with null columns and writes a
warning in the log.

In these cases, a note is written in the log: “WARNING: A table has been extended with null columns
to perform the UNION set operation.”

Behavior of UNION Operator with ALL

proc sql;
select CustomerID
from sq.salesemail The ALL keyword
union all
select CustomerID does not remove
from sq.salesphone the duplicate rows.
order by CustomerID;
quit;

The ALL keyword does not remove the duplicate rows.

ALL Keyword Considerations

Us e when duplicates in Us e when

fi nal result will not duplicates a re not
ca use problems possible
Using the ALL keyword improves efficiency of the set operators because SAS
does not make a second pass to remove duplicates.

The ALL keyword is applicable only in the INTERSECT, UNION, and EXCEPT set operators.

Combining Set Operators

saleslist
Create a list of
EXCEPT customers who have
not responded to
either email or phone
UNION
sales attempts.

salesemail

salesphone

Combining Set Operators

proc sql; Find a unique list of customers who

select CustomerID have responded using an in-line view.
from sq.saleslist
except
(select customerid
from sq.salesemail
union
select customerID
from sq.salesphone);
quit;

Combining Set Operators

proc sql;
select CustomerID
from sq.saleslist
except
(select customerid
from sq.salesemail
union
select customerID
from sq.salesphone);
quit;
We have two customers who
have not responded to any of
our sales attempts.

Syntax Summary

SELECT query…
UNION | EXCEPT | INTERSECT <ALL> <CORR>
SELECT query…;
Set Operators

Practice

Level 1
1. Using the EXCEPT SET Operator
Use the EXCEPT SET operator to generate a report listing of merchants in the sq.merchant
table who are not listed in the sq.transaction table.
a. Select MerchantID from the sq.merchant table.
b. Use the EXCEPT SET operator.
c. Select MerchantID from the sq.transaction table.
d. Order the results by MerchantID.
e. Add an appropriate title.
Results

f. How many merchants do not have a transaction?

2. Using the UNION SET Operator
Using the sq.employee table as input, create a report that displays the total salary and total
number of employees. In addition, add rows to display the salary information for US and AU
employees separately. Use the UNION SET operator to combine the three rows of output for the
final report.
a. Write a query to create a report for all employees.
1) Create the first row of the report. Include the constant text Total Paid to All Employees,
the sum of Salary formatted with the DOLLAR. format, and the total number of
employees using the COUNT function. Name the count c olumn Total.
2) Run the query and compare your results.
Results

b. Write a query to create a report for all US employees.

1) Create the second row of the report. Include the constant text Total Paid to US
Employees, the sum of Salary formatted with the DOLLAR. format, and the total number
employees using the COUNT function. Name the count column Total.
2) Filter the data for US employees by using the UPCASE function to select rows with the
Country value US.
3) Run the query and compare your results.
Results

c. Write another query to create the last row of the report for all AU employees.
1) Follow the pervious steps. Replace US with AU.
2) Run the query and compare your results.
Results

d. Combine the results of all three queries into a single query using UNION SET operators.
1) Order the results by descending total salary.
2) Add an appropriate title.
Results

e. Which country has more employees?

Level 2
3. Using the EXCEPT SET Operator with the DISTINCT Keyword
Using the sq.statepopulation table, generate a list of state codes for states without any
customers.
a. Write a query to list the unique Name values in the sq.statepopulation table. This list
represents all available states.

b. Write a separate query to list the unique State values from the sq.customer table. This list
represents all states where customers reside.
c. Combine the queries using the EXCEPT SET operator to display states with no customers.
Add an appropriate title.
Results

d. For which value (or values) of State are there no customers?

Challenge
4. Using SET Operators to Summarize Data
Determine what percentage of customers have accepted either the phone or email offer. The
sq.saleslist table contains the full list of customers presented with an offer. The sq.salesemail
and sq.salesphone tables contain email and phone responses.
a. Write a query to use a UNION SET operation to combine the CustomerID values for
customers who accepted either the phone or email offer. This will form the basis of your in-
line view.
Results

b. Write a query to count the number of customers who have accepted either offer (step a
above).
1) Use the following formula to calculate the rate of offer acceptance:
select count(*)/(select count(*) from sq.saleslist)
from (your in-line view code from step 1)
2) Name the calculated column PctResp. Format it as a percent with no decimals.
3) Label the new column Offer Acceptance Rate.
4) Add an appropriate title to the report.

Results

c. What is the acceptance rate?

5.3 OUTER UNION

OUTER UNION Operator

Desired Results
salesemail

Create a report that

contains a list of all
customer phone and
email responses.
salesphone

OUTER UNION Operator

SELECT query… CustomerID EmailResp

OUTER UNION <CORR> 1905044343 Declined
SELECT query…; 1908694347 Accepted
… …
proc sql; Result set 1
select *
from sq.salesemail
outer union
select * CustomerID SalesRep PhoneResp
from sq.salesphone;
1939774314 121038 Declined
quit;
1963960449 120145 Declined
Result set 2 … … …

Default Behavior of the OUTER UNION Operator

ALL columns and rows are

combined. Same name
columns are not overlaid.

Using the OUTER UNION Set Operator

Scenario
Use the OUTER UNION set operator with the CORR keyword to combine the salesemail and
salesphone tables.

Files
• s105d02.sas
• salesemail – a SAS table that contains email response information from targeted customers
• salesphone – a SAS table that contains phone response information from targeted customers

Syntax

PROC SQL;
SELECT query…
OUTER UNION <CORR>
SELECT query…;
QUIT;

table(RENAME=(old-name-1=new-name-1 …))

Notes
• The OUTER UNION set operator produces all unique rows from both queries.
• The CORR keyword overlays columns that have the same name in both tables. When used with
OUTER UNION, CORR includes columns that are not in both tables
• The RENAME= data set option changes the name of a column.

Demo
1. Open the s105d02.sas program in the demos folder and find the Demo section. Run the query
to perform an OUTER UNION concatenation of the sq.salesemail and sq.salesphone tables.
Examine the results.
proc sql;
select *
from sq.salesemail
outer union
select *
from sq.salesphone;
quit;
2. Add the CORR keyword after the OUTER UNION set operator. Run the query and examine the
results.
proc sql;
select *
from sq.salesemail
outer union corr

select *
from sq.salesphone;
quit;
3. Add the RENAME= option after the salesemail table and rename the column EmailResp to
Resp. After the salesphone table, rename the column PhoneResp to Resp. Run the query and
examine the results.
proc sql;
select *
from sq.salesemail(rename=(EmailResp=Resp))
outer union corr
select *
from sq.salesphone(rename=(PhoneResp=Resp));
quit;
4. Remove the RENAME= option after each table. Modify the first SELECT clause and select the
column CustomerID. In the first clause, also select EmailResp and change the column name to
Resp using the AS keyword. Modify the second SELECT clause and select the CustomerID,
SalesRep, and PhoneResp columns. Change the PhoneResp column name to Resp using the
AS keyword. Run the query and examine the results.
proc sql;
select CustomerID, EmailResp as Resp
from sq.salesemail(rename=(EmailResp=Resp))
outer union corr
select CustomerID, SalesRep, PhoneResp as Resp
from sq.salesphone(rename=(EmailResp=Resp));
quit;
5. Add the CREATE TABLE statement to create a table from the query results. Name the table
response1.
proc sql;
create table response1 as
select CustomerID, EmailResp as Resp
from sq.salesemail
outer union corr
select CustomerID, SalesRep, PhoneResp as Resp
from sq.salesphone;
quit;
6. Find the SAS DATA Step section. Briefly describe how you can retrieve the same results using
the SAS DATA step.
data response2;
length Resp $12;
set sq.salesemail(rename=(EmailResp=Resp))
sq.salesphone(rename=(PhoneResp=Resp));
run;

SQL Versus Traditional SAS Programming

proc sql;
create table response1 as
select CustomerID, EmailResp as Resp
from sq.salesemail
outer union corr
select CustomerID, SalesRep,
PhoneResp as Resp
from sq.salesphone;
quit;
data response2;
length Resp $12;
set sq.salesemail(rename=(EmailResp=Resp))
sq.salesphone(rename=(PhoneResp=Resp));
run;

The OUTER UNION set operator is similar to the DATA step with multiple tables listed in the SET
statement. OUTER UNION in SQL and the DATA step with the SET statement can produce similar
results.

Syntax Summary

SELECT query…
OUTER UNION <CORR>
SELECT query…;
OUTER UNION

table(RENAME=(old-name-1=new-name-1 …))
RENAME Data Set Option

Beyond SQL Essentials

What if you want to ...

. . . l ea rn more a bout . . . l ea rn more a bout . . . l ea rn more a bout

combi ni ng a nd us i ng SET operators ? the SQL procedure?
modi fyi ng SAS da ta
s ets ?
• Read Combining and • Read SQL Set Operators: So • Visit the SQL Procedure
Modifying SAS Data Sets: Handy Venn You Need Them. section on the ELP for a list
Examples, Second Edition. of SAS papers.
• View Combining SAS Data
Sets: Methods in the SAS
documentation

Practice

Level 1
5. Using the OUTER UNION SET Operator
Create a report that shows the email and phone offer responses along with the sales
representative if available.
a. Using the sq.salesphone table as input, write a query to list the following columns:
1) CustomerID
2) a new column named Response based on the existing PhoneResp column
3) SalesRep labeled Sales Rep
4) a new column named Channel with the constant text Phone
Run the query and compare your results.
Results

b. Using the sq.salesemail table as input, write a query to list the following columns:
1) CustomerID
2) a new column named Response based on the existing EmailResp column
3) a new column named Channel with the constant text Email
Run the query and compare your results.
Results

c. Combine the two query results using the OUTER UNION SET operation.
1) Be mindful of the column alignment. Use the SET operator modifiers as needed.
2) Order the results by CustomerID and Response.
3) Add an appropriate title.
4) Run the query to generate the final results.
Results

d. To how many sales representatives can we attribute an accepted offer?

5.4 Solutions
Solutions to Practices
1. Using the EXCEPT SET Operator
/*s105s01.sas*/

title 'Merchants without Transactions';

proc sql;
select MerchantID
from sq.merchant
except
select MerchantID
from sq.transaction
order by MerchantID;
quit;
title;
How many merchants do not have a transaction? Six merchants do not have a transaction.
2. Using the UNION SET Operator
/*s105s02.sas*/

/*a*/
proc sql;
select 'Total Paid to All Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee;
quit;

/*b*/
proc sql;
select 'Total Paid to US Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee
where upcase(Country)='US';
quit;

/*c*/
proc sql;
select 'Total Paid to AU Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee
where upcase(Country)='AU';
quit;

/*d*/
title 'Country Specific Salary Information';
proc sql;
select 'Total Paid to All Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee
union
select 'Total Paid to US Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee
where upcase(Country)='US'
union
select 'Total Paid to AU Employees',
sum(Salary) format=dollar16.,
count(*) as Total
from sq.employee
where upcase(Country)='AU'
order by 2 desc;
quit;
title;
Which country has more employees? The US has more employees.
3. Using the EXCEPT SET Operator with the Distinct Keyword
/*s105s03.sas*/

title 'States with No Customers';

proc sql;
select Name
from sq.statepopulation
except
select distinct State
from sq.customer;
quit;
title;
For which value (or values) of State are there no customers?
PR (Puerto Rico) contains no customers.
4. Using SET Operators to Summarize Data
/*s105s04.sas*/

title 'Acceptance Rate';

proc sql;
select count(*)/(select count(*)
from sq.saleslist) as PctResp format=percent5.
label='Offer Acceptance
Rate'
from(select CustomerID

from sq.salesemail
where EmailResp = 'Accepted'
union
select CustomerID
from sq.salesphone
where PhoneResp = 'Accepted');
quit;
title;
What is the acceptance rate? 50%
5. Using the OUTER UNION Set Operator
/*s105s05.sas*/

title 'Offer Results with Sales Rep';

proc sql;
select CustomerID,
PhoneResp as Response,
SalesRep 'Sales Rep',
'Phone' as Channel
from sq.salesphone
outer union corr
select CustomerID,
EmailResp as Response,
'Email' as Channel
from sq.salesemail
order by 1,2;
quit;
title;
To how many sales representatives can we attribute an accepted offer?
Two sales representatives

Solutions to Activities and Questions

continued...
5.01 Activity – Correct Answer
2. In the Intersect s ecti on, exa mine a nd run the query. Di d the query run
s ucces sfully? Why not? No. By default, the INTERSECT set operator
matches by column position, and the data must be of the same type.

proc sql; EmailResp is

select *
from sq.salesemail character. SalesRep is
intersect numeric
select *
from sq.salesphone;
quit;

5.01 Activity – Correct Answer

3. Add the CORR opti on a fter the INTERSECT keyword. Di d the query run
s ucces sfully? Why? Yes. The CORR keyword matches columns by name,
and nonmatching columns are removed from the intermediate result
sets.

proc sql;
select * CORR matches by
from sq.salesemail column names and
intersect corr removes
suppresses
columns
columnsnot
select * not
in in
both
both
tables.
tables
from sq.salesphone;
quit;

5.02 Activity – Correct Answer

select CustomerID select *

from sq.saleslist from sq.saleslist
except except corr
select CustomerID select *
from sq.salesphone; from sq.salesphone;

Five customers

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
Lesson 6 Using and Creating
Macro Variables in SQL
6.1 Creating User-Defined Macro Variables........................................................................ 6-3

6.2 Creating Data-Driven Macro Variables with PROC SQL ................................................ 6-6
Demonstration: Using a PROC SQL Data-Driven Macro Variable ................................ 6-10
Demonstration: Concatenating Values in Macro Variables ........................................... 6-17
Practice............................................................................................................... 6-22

6.3 Solutions ................................................................................................................... 6-28

Solutions to Practices ............................................................................................ 6-28
Solutions to Activities and Questions........................................................................ 6-33
6-2 Lesson 6 Using and Creating Macro Variables in SQL

6.1 Creating User-Defined Macro

Variables

Efficiently Changing the Filter Value

proc sql;
create table CustomerGA as
select CustomerID, Employed, Race,
Married, State, CreditScore
from sq.customer
where State="GA" and To change to the values
CreditScore > 650; to NC and 700, you
quit; would have to manually
adjust the query.

GA  NC 650  700

Efficiently Changing the Filter Value

NY WV
A SAS macro variable
stores text that is
ma cro substituted in your code
va ri a ble when it runs. It’s like
700
an automatic
775 find-and-replace.
725

ma cro
va ri a ble

Creating User-Defined Macro Variables

create
%LET macro-variable=value; the
macro
%let State=GA; variables
%let CreditMin=650;

proc sql; Macro Variables

create table CustomerGA as Name Valu
select CustomerID, Employed, Race,
Married, State, CreditScore e
from sq.customer State GA
where State="GA" and
CreditScore > 650; CreditMin 650
quit;
Macro variables are temporary, so when you exit
SAS, they are deleted.
5
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Macro language statements begin with a % sign. To create a user-defined macro variable, you use
the %LET statement. In the %LET statement, you specify the name of the macro variable, followed
by an equal sign and then the text string that you want to store.
It is recommended that you do not include quotation marks when you define the macro variable
value. Use quotation marks when necessary after the macro variable is resolved.

Using User-Defined Macro Variables

&macro-variable; use the

macro
%let State=GA; variables
%let CreditMin=650;

proc sql;
create table Customer&State as When the query executes,
select CustomerID, Employed, Race, values of State and
Married, State, CreditScore CreditMin are replaced by
from sq.customer GA and 650.
where State="&State" and
CreditScore > &CreditMin;
quit;

Macro variables must be enclosed in double quotation marks when the macro variables are used as
a character string.

6.01 Activity
Open s106a01.sas from the activities fol der a nd perform the following tasks
to crea te a nd us e user-defined macro va ri ables:
1. Run the progra m. Exa mi ne the log a nd results . Confirm tha t the name of
the newl y crea ted table i s customerga a nd contains 957 rows .
2. Repl ace the va lues GA a nd 650 i n the %LET s tatements wi th NC a nd 700.
Run the progra m. Exa mi ne the log a nd results . What i s the name of the
newl y crea ted table? How ma ny rows ?
3. Cha nge the double quotation marks i n the WHERE cl a use expression to
s i ngle quotation ma rks. Run the query. How ma ny rows a re i n the new
ta bl e?

6.2 Creating Data-Driven Macro Variables

with PROC SQL

Creating Data-Driven Macro Variables

Create macro
PROC SQL Query variables from values
returned by a query.

In PROC SQL, you can create macro variables from values returned by a query. You can then
reference those stored macro variables in other PROC steps, DATA steps, titles, and footnotes.
Think of it this way: Build once, use many.

Subquery Returning a Value

select Name, PopEstimate1

from sq.statepopulation
where PopEstimate1 > (select avg(PopEstimate1)
from sq.statepopulation);

How can we
Avera ge Es ti mated Population for Next Yea r i s: ??? document the
average value of
PopEstimate1 in the
title?

Steps for Creating Data-Driven Macro Variables

1 2

Create the Macro Variable Use the Macro Variable

Using PROC SQL

Step 1: Creating the Macro Variable Using PROC SQL

SELECT col-name
INTO :macvar
FROM table;
select avg(PopEstimate1)
The value is stored in the
into :AvgEst1
from sq.statepopulation; AvgEst1 macro variable.

Macro Variables
Name Value
AvgEst1 6278420

If the query produces more than one row of output, the macro variable will contain only the value
from the first row.
If the query has no rows in its output, then the macro variable is not modified. If the macro variable
does not exist yet, it will not be created.

Displaying Macro Variable Values

%PUT text; %PUT displays the resolved macro value

in the SAS log.
%put AvgEst is &AvgEst1;

%put &=AvgEst1;

%put AvgEst is &AvgEst1;

AvgEst is 6278420

%put &=AvgEst1;
AVGEST1= 6278420

A %PUT statement that contains &= followed by the name of the macro variable displays the name
and value of a macro variable.

Step 2: Using the Macro Variable

title "Average Estimated Population for Next Year: &AvgEst1";

proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > &AvgEst1;
quit; Now we can use the
macro variable AvgEst1 in
titles and queries.

Using a PROC SQL Data-Driven Macro Variable

Scenario
Use PROC SQL to create and use a macro variable.

Files
• s106d01.sas
• statepopulation – a SAS table that contains estimated yearly population by state

Syntax

PROC SQL NOPRINT;

SELECT col-name
INTO :macvar
FROM table;
QUIT;

%PUT &=macvar;
OPTIONS SYMBOLGEN;

Notes
• A macro variable stores a text string that can be substituted into a SAS program.
• If the query produces more than one row of output, the macro variable will contain only the value
from the first row.
• Macro variables can be referenced in a program by preceding the macro variable name with
& (ampersand).
• The %PUT statement writes the value of the macro variable to the SAS log.
• The SYMBOLGEN option displays a note identifying the macro variable and its resolved value
in the log.

Demo
1. Open the s106d01.sas program in the demos folder and find the Demo section. Run the first
query in the Create the Macro Variable Using PROC SQL section. View the results.
proc sql;
select avg(PopEstimate1)
into :AvgEst1
from sq.statepopulation;
quit;

2. In the same query, add the NOPRINT option in the PROC SQL statement. At the end of the
query, add a %PUT statement to view the value of the newly created macro variable in the SAS
log. Run the query and view the SAS log.
proc sql noprint;
select avg(PopEstimate1)
into :AvgEst1
from sq.statepopulation;
quit;
%put &=AvgEst1;

3. Move to STEP 2 and apply the newly created macro variable AvgEst1 in the TITLE statement
and the WHERE clause. Run the query. View the log and results.
Note: To format a macro variable in the TITLE statement, you can use
%left(%qsysfunc(putn(&AvgEst1,dollar16.))). Scroll to the bottom of the program file for
this demo to view the solution code.
title "Average Estimated Population for Next Year: &AvgEst1";
proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > &AvgEst1;
quit;
4. Above the query, add the statement OPTIONS SYMBOLGEN to see the value that was
substituted in the code. Below the query, turn of the options by adding the OPTIONS
NOSYMBOLGEN statement. Run the query and examine the log.
options symbolgen;

…sql query

options nosymbolgen;

Displaying Macro Variable Values

Use the SYMBOLGEN system

OPTIONS SYMBOLGEN | NOSYMBOLGEN; option to see the value that was
substituted in the code.
options symbolgen;
title "Average Estimated Population for Next Year: &AvgEst1";
proc sql;
select Name, PopEstimate1
from sq.statepopulation
where PopEstimate1 > &AvgEst1;
quit;
options nosymbolgen;
31 o pt i on s sy mb ol ge n;
SY MB OL GE N: Ma c ro v ar ia bl e AV GE S T1 r es ol ve s to 6 27 84 20
32 t it l e "A ve ra ge E st im a te d Po pu la ti on fo r Ne xt Y ea r: & A vg Es t1 ";
33 p ro c s ql ;

The SYMBOLGEN option displays the results of resolving macro variable references in the SAS log.
This option is useful for debugging.

Creating Multiple Macro Variables

proc sql noprint;
select avg(PopEstimate1), min(PopEstimate1),
max(PopEstimate1), count(PopEstimate1)
into :AvgEst1, :MinEst1, :MaxEst1, :TotalCount
from sq.statepopulation;
quit;

Macro Variables
The INTO clause can create multiple Name Value
macro variables from query results. AvgEst1 6278420
MinEst1 584290
MaxEst1 39209127
TotalCount 52
20
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

You list the names of the macro variables to be created in the INTO clause. Precede each macro
variable name with a colon. PROC SQL generates the values that are assigned to these variables by
executing the query.

Displaying Macro Variables

Display each new macro variable and
its value in the log.
%put &=AvgEst1;
%put &=TotalCount;
%put Min Value is &MinEst1;
%put Max: &MaxEst1;
Do you notice anything
interesting about the
37 %put &=AvgEst1;
AVGEST1= 6278420 way the values are
38 %put &=TotalCount; displayed?
TOTALCOUNT= 52
39 %put Min Value is &MinEst1;
Min Value is 584290
40 %put Max: &MaxEst1;
Max: 39209127

Displaying Macro Variables

By default, macro variables

that contain numeric values
are formatted using the
BEST8. format.
37 %put &=AvgEst1;
AVGEST1= 6278420
38 %put &=TotalCount;
TOTALCOUNT= 52
39 %put Min Value is &MinEst1;
Min Value is 584290
40 %put Max: &MaxEst1;
Max: 39209127

When storing a value in a single macro variable, PROC SQL preserves leading or trailing blanks. By
default, when creating macro variables that contain numeric values, the values are formatted using
the BEST8. format, which preserves leading blanks for numbers that have fewer than eight digits.
This can also cause the value to be displayed using scientific notation for numbers that have more
than eight digits. You can use another format, such as the w. format, to display the value without
using scientific notation.

Removing Leading and Trailing Blanks

INTO :macvar TRIMMED The TRIMMED option can be used to trim the
leading and trailing blanks.
proc sql noprint;
select avg(PopEstimate1), min(PopEstimate1),
max(PopEstimate1), count(PopEstimate1)
into :AvgEst1 trimmed, :MinEst1 trimmed,
:MaxEst1 trimmed, :TotalCount trimmed
from sq.statepopulation;
quit; 48 %put &=AvgEst1;
%put &=AvgEst1; AVGEST1=6278420
%put &=TotalCount; 49 %put &=TotalCount;
TOTALCOUNT=52
%put Min Value is &MinEst1; 50 %put Min Value is &MinEst1;
%put Max: &MaxEst1; Min Value is 584290
51 %put Max: &MaxEst1;
Max: 39209127

6.02 Activity
Open s106a02.sas from the activities fol der a nd perform the following tasks
to s tore a number l arger tha n eight di gits i n a macro va ri able:
1. Run the progra m i n STEP 1 to crea te the ma cro va ri ables MaxPop a nd
TotalCtry. The ma cro va ri a bles s tore the estimated ma ximum three-year
popul ation estimate for a country a nd the tota l number of countri es.
Vi ew the l og. Notice tha t MaxPop s tores a va lue i n s cientific notation.
2. Exa mi ne STEP 2 of the progra m to fi nd the country wi th the ma xi mum
es timated three-year population. Run the program to us e the ma cro
va ri a bles tha t you crea ted. Confirm tha t no rows were returned.
3. Add the FORMAT=10. col umn modifier to the max(estYear3Pop) col umn
i n STEP 1. Run the entire program. Whi ch country ha s the l argest three -
yea r es timated population?
24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Subquery Returning Multiple Values

proc sql;
create table division3 as
select *
from sq.customer
where State in (select Name
from sq.statepopulation
where Division = '3');
quit;
How can we store a
"IL","IN","MI","OH","WI" list of values in a
macro variable?

Concatenating Values in Macro Variables

SELECT col-name
INTO :macvar SEPARATED BY "delimiter"
FROM table;

proc sql noprint;

select Name
into :StateList separated by ","
from sq.statepopulation
where Division = '3';
quit;

%put &=StateList;
48 %put &=StateList;
STATELIST=IL,IN,MI,OH,WI

Concatenating Values in Macro Variables

title "States in Division 3: &StateList";

proc sql;
create table division3 as
select *
from sq.customer
where State in (&StateList);
quit; Will you be able to use the
StateList macro variable to
subset the table?
48 %put &=StateList;
STATELIST=IL,IN,MI,OH,WI

Concatenating Values in Macro Variables

title "States in Division 3: &StateList";

proc sql;
create table division3 as
select *
from sq.customer
where State in (IL,IN,MI,OH,WI);
quit; Character values need to
each be enclosed in
quotation marks.

Concatenating Values in Macro Variables

Scenario
Use PROC SQL to create a concatenated list of state values separated by a comma, with each
value enclosed in quotation marks.

Files
• s106d02.sas
• statepopulation – a SAS table that contains estimated yearly population by state
• customer – a SAS table that contains one row per customer

Syntax

PROC SQL;
SELECT col-name
INTO :macvar SEPARATED BY "delimiter"
FROM table;
QUIT;

QUOTE(argument)
STRIP(argument)

Notes
• You can concatenate the values of one column into one macro variable.
• Use the SEPARATED BY keywords to specify a character to delimit the values in the macro
variable.
• The QUOTE function adds double quotation marks to a character value.
• The STRIP function returns a character string with all leading and trailing blanks removed.

Demo
1. Open the s106d02.sas program in the demos folder and find the Demo section. Run the code
in Step 1: Create Macro Variables. Examine the results and log to see the values of the newly
created macro variables &Division and &StateList.
%let Division=3;

proc sql;
select Name
into :StateList SEPARATED BY ","
from sq.statepopulation
where Division = "&Division";
quit;
%put &=Division;
%put &=StateList;

2. Discuss the query under Step 2: Use Macro Variables. The table being created will end with the
value of the macro variable &Division. The customer table will attempt to be subset by a list of
values from the &StateList macro variable. Run the query and examine the errors.
Note: The name of the new table will end with the value of the macro variable.
options symbolgen;
proc sql;
create table division&Division as
select *
from sq.customer
where State in (&StateList);
quit;
options nosymbolgen;
3. Move back to Step 1. Add the QUOTE function around the Name column. Run the query
and %PUT statements. Examine the results and log.
proc sql;
select quote(Name)
into :StateList SEPARATED BY ","
from sq.statepopulation
where Division = "&Division";
quit;
%put &=Division;
%put &=StateList;
4. Add the STRIP function inside the QUOTE function. Run the query and %PUT statements.
Examine the results and log.
proc sql;
select quote(strip(Name))
into :StateList SEPARATED BY ","
from sq.statepopulation
where Division = "&Division";
quit;
%put &=Division;
%put &=StateList;
5. Move to Step 2. Run the query. Examine the results and log.
options symbolgen;
proc sql;
create table division&Division as
select *
from sq.customer
where State in (&StateList);
quit;
options nosymbolgen;

6. Move to Step 1. Change the value in the %LET statement to 9 and add the NOPRINT option
in the PROC SQL statement to finalize the program. Run the entire program. Examine the
results and log.
%let Division=9;

proc sql noprint;

…

Using Formats when Creating Macro Variables

FORMAT=$QUOTEw.

proc sql noprint;

select Name format=$quote4.
into :StateList SEPARATED BY ","
from sq.statepopulation
where Division = '3'; To ensure that values are not truncated,
quit; you should specify a width as long as the
longest value +2.
%put &=StateList;

STATELIST="IL","IN","MI","OH","WI"

Strings larger than six characters are truncated when you use the QUOTE format. To ensure that
values are not truncated, you should be sure to specify a width as long as the longest value +2.
When using other formats, be sure to check the SAS documentation.

Syntax Summary

SELECT col-name, col-name

INTO :macvar1, :macvar2 <TRIMMED>
FROM table;
Create Data-Driven Macro Variables

SELECT col-name PROC SQL NOPRINT;

INTO :macvar1 SEPARATED BY "delimiter" Suppress Display
FROM table;
Concatenate Macro Values
%PUT text &=macvar;
OPTIONS SYMBOLGEN | NOSYMBOLGEN Display Macro Value
Display Resolved Macro Variables
35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Beyond SQL Essentials

What if you want to ...

. . . l ea rn more a bout . . . l ea rn more a bout . . . l earn more about

SAS functi ons the SAS ma cro us i ng the SAS macro
l a nguage l a nguage and SQL

• Take the SAS Functions by • Take the SAS Macro Language • Visit the SAS Macro
Example course. 1: Essentials course. Language section on the ELP
• Read SAS Functions by • Read SAS Macro Programming for a list of SAS papers and
Example. Made Easy. blogs.
• View the SAS Function • View the SAS 9.4 Macro
Documentation. Language: Reference
• Create your own functions Documentation.
with PROC FCMP.

Practice

Level 1
1. Creating a Macro Variable from an SQL Query
Using the sq.statepopulation table, write a program that dynamically returns states in a
specified region that have a three-year estimated population change greater than the median
population change for the entire region.
a. Write a query to create the macro variable MedianEst to store the value of the median
nPopChg3 of Region 1.
1) Calculate the median of nPopChg3.
2) Use the INTO clause to create a macro variable named MedianEst. Use the TRIMMED
keyword.
3) Use the sq.statepopulation table.
4) Filter the results for rows in the Region column with the value of 1. The value 1 is
character.
5) Add the NOPRINT option in the PROC SQL statement.
6) Below the query view the new macro variable in the log using %PUT.
%put &=MedianEst;
7) Run the query and compare your log.

b. Write another query to select states and their three-year population change in Region 1 that
have a higher estimate than the median of all states in that region.
1) Select the Name and nPopChg3 columns. Format nPopChg3 so that values are
displayed with commas.
2) Use the sq.statepopulation table.
3) Filter rows for states in region 1 that have an nPopChg3 value greater than the macro
variable &MedianEst created in step 1.
4) Order by nPopChg3 descending.
5) Add the following titles:
a) title 1 – States in Region 1 with a 3-Year Estimated Population Change Greater
than the Median
b) title 2 – Median Estimate: &MedianEst
6) Run the query and compare your results.

Results

c. At the beginning of your program, create a macro variable named RegionNum to specify the
region to analyze.
1) Use the %LET statement to create the macro variable RegionNum and set the value
equal to 1.
2) In you program, replace every location of the character value 1 with &RegionNum. Make
sure that the macro variable is enclosed in double quotation marks.
Hint: You need to do this replacement in the WHERE clause in each query and in the
TITLE statement.
3) Run the entire program and confirm that the results are the same as step b.
d. In the %LET statement replace the value 1 with the value 2. Run the entire program.
Results

e. How many states in Region 2 have a higher three-year estimated population change than the
median of all states in that region?

Level 2
2. Creating a Macro Variable with a List of Values from an SQL Query
Write a program that dynamically determines a list of countries in a specified region. Then
determine the percentage of people who have borrowed from a financial institution or used a
credit card (% age 15+), and whether it is increasing or decreasing from the year 1 estimate to
the year 3 estimate. Use the sq.globalmetadata table for country region information and the
sq.globalfindex table for country financial and population information.

a. Using sq.globalmetadata, write a query to create a macro variable that lists the distinct
values of CountryCode for a specified region. Ensure that the values are enclosed in
quotation marks and separated by commas.
1) Create a macro variable named RegionValue with the value South Asia.
2) Use the RegionValue macro variable in a query to create a data-driven macro variable
named Countries that lists the country codes in the specified region. The value list
should be comma separated, and each value should be enclosed in quotation marks.
3) Use the NOPRINT option in the PROC SQL statement.
4) Use %PUT below your query to view the value of the macro variable Countries.
5) Run the query and %PUT statement. Compare your results.
Results

b. Write another query to select the countries in the region that you specified. Convert the
whole numbers to percentages, and determine whether the estimated value is increasing,
decreasing, or unknown using the sq.globalfindex table.
1) Select the CountryCode and IndicatorName columns. Create three new columns.
a) First divide EstYear1 by 100 and name the column EstYear1PCT. Format using
percentages.
b) Follow the previous step for EstYear3.
c) Use the CASE expression to determine whether the value is Increasing, Decreasing,
or Unk nown. The value is Unk nown when the value is null. Name the column
Forecast.
Hint: In the CASE expression, make sure that the first WHEN tests whether a value
is missing.
2) Filter rows by CountryCode using the Countries macro variable and whether
IndicatorName is equal to Borrowed from a financial institution or used a credit card
(% age 15+).
3) Order by Forecast.
4) Add the title Countries in &RegionValue.
5) Run the query and compare the results.

Results

c. Change the RegionValue macro variable at the beginning of the program to North America.
Run the program.
Results

d. What is the forecast when CountryCode is CAN?

3. Creating Dynamic Titles with Formats
Write a program that dynamically determines the total number of customers in specific states
and uses that value to calculate the percentage of customers in a ZIP code in a specific state.
Use the sq.customer table.
a. Write a query using a macro variable to count the total number of customers in a specified
state.
1) Use the %LET statement to create a macro variable named StateValue. Set the value
equal to NC.
2) Use the macro variable StateValue to subset the table and count the number of
customers in that state. Store the number in a macro variable named TotalCust. Use the
TRIMMED option.
3) Use the NOPRINT option and view the value of TotalCust in the log using %PUT.
4) Run the program and compare the results.

b. Write another query to calculate the percentage of customers in each ZIP code for the
specified state.
1) Select Zip and calculate the percentage of customers in a ZIP code by counting the
number of customers with that Zip value and dividing it by the TotalCust macro variable.
Name the new column PctZip and format it using a percent.
2) Filter the rows by State using the macro variable StateValue.

3) Group the results by Zip.

4) Order by PctZip descending.
5) Add dynamic titles using the following:
title "Total Customers in &StateValue:
%left(%qsysfunc(putn(&TotalCust,comma15.)))";
title2 "Percentage of Customers by Zip";
title3 "Report Created on %left(%qsysfunc(today(),weekdate.))";
6) Run the program and compare the results.
Partial Results

c. Change the value in the macro variable State from NC to TX. Run the program and compare
the results.
Partial Results

d. In Texas, which Zip value has the highest percentage of customers?

Challenge
4. Splitting One Table into Many
Write a program that dynamically creates a new table for each distinct value in a column. Use
the Region column in the sq.globalmetadata table.
Hint: Visit the Extended Learning page and view the SAS blog How to split one data set into
many in the SAS Macro Language section or follow the direct link:
https://blogs.sas.com/content/sasdummy/2015/01/26/how-to-split-one-data-set-into-many/
How many tables were created?

6.3 Solutions
Solutions to Practices
1. Creating a Macro Variable from an SQL Query
/*s106s01.sas*/

/*a*/
proc sql noprint;
select median(nPopChg3)
into :MedianEst trimmed
from sq.statepopulation
where Region="1";
quit;
%put &=MedianEst;

/*b*/
title "States in Region 1 with a 3-Year Estimated Population
Change Greater than the Median";
title2 "Median Estimate: &MedianEst";
proc sql;
select Name,nPopChg3 format=comma14.
from sq.statepopulation
where Region="1" and
nPopChg3 > &MedianEst
order by nPopChg3 desc;
quit;
title;

/*c*/
%let RegionNum=2;

proc sql noprint;

select median(nPopChg3)
into :MedianEst trimmed
from sq.statepopulation
where Region="&RegionNum";
quit;
%put &=MedianEst;

title "States in Region &RegionNum with a 3 -Year Estimated

Population Change Greater than the Median";
title2 "Median Estimate: &MedianEst";
proc sql;
select Name,nPopChg3 format=comma14.
from sq.statepopulation
where Region="&RegionNum" and
nPopChg3 > &MedianEst

order by nPopChg3 desc;

quit;
title;
How many states in Region 2 have a higher three-year estimated population change than the
median of all states in that region? Six
2. Creating a Macro Variable with a List of Values from an SQL Query
/*s106s02.sas*/

/*a*/
%let RegionValue=South Asia;
proc sql noprint;
select quote(strip(CountryCode))
into :Countries separated by ","
from sq.globalmetadata
where Region="&RegionValue";
quit;
%put &=Countries;

/*b*/
title "Countries in &Region";
proc sql;
select CountryCode, IndicatorName,
EstYear1/100 as EstYear1PCT format=percent7.2,
EstYear3/100 as EstYear3PCT format=percent7.2,
case
when (EstYear1 is null or EstYear3 is null) then "Unknown"
when calculated EstYear1PCT < calculated EstYear3PCT
then "Increasing"
when calculated EstYear1PCT > calculated EstYear3PCT
then "Decreasing"
else "No Change"
end as Forecast
from sq.globalfindex
where CountryCode in (&Countries) and
IndicatorName="Borrowed from a financial institution or
used a credit card (% age 15+)"
order by Forecast;
quit;
title;

/*c*/
%let Region=North America;
proc sql noprint;
select quote(strip(CountryCode))
into :Countries separated by ","
from sq.globalmetadata
where Region="&Region";
quit;

%put &=Countries;

title "Countries in &Region";

proc sql;
select CountryCode, IndicatorName,
EstYear1/100 as EstYear1PCT format=percent7.2,
EstYear3/100 as EstYear3PCT format=percent7.2,
case
when (EstYear1 is null or EstYear3 is null) then "Unknown"
when calculated EstYear1PCT < calculated EstYear3PCT
then "Increasing"
when calculated EstYear1PCT > calculated EstYear3PCT
then "Decreasing"
else "No Change"
end as Forecast
from sq.globalfindex
where CountryCode in (&Countries) and
IndicatorName="Borrowed from a financial institution or
used a credit card (% age 15+)"
order by Forecast;
quit;
title;
What is the forecast when CountryCode is CAN? Increasing
3. Creating Dynamic Titles with Formats
/*s106s03.sas*/

/*a*/
%let StateValue=NC;
proc sql noprint;
select count(*) as Total
into :TotalCust trimmed
from sq.customer
where State="&StateValue";
quit;
%put &=TotalCust;

/*b*/
title "Total Customers in &StateValue:
%left(%qsysfunc(putn(&TotalCust,comma15.)))";
title2 "Percentage of Customers by Zip";
title3 "Report Created on %left(%qsysfunc(today(),weekdate.))";
proc sql;
select Zip, count(*)/&TotalCust as PctZip format=percent8.2
from sq.customer
where State="&StateValue"
group by Zip
order by PctZip desc;
quit;

title;

/*c*/
%let StateValue=TX;

proc sql noprint;

select count(*) as Total
into :TotalCust trimmed
from sq.customer
where State="&StateValue";
quit;
%put &=TotalCust;

title "Total Customers in &StateValue:

%left(%qsysfunc(putn(&TotalCust,comma15.))) ";
title2 "Percentage of Customers by Zip";
title3 "Report Created on %left(%qsysfunc(today(),weekdate.)) ";
proc sql;
select Zip, count(*)/&TotalCust as PctZip format=percent8.2
from sq.customer
where State="&StateValue"
group by Zip
order by PctZip desc;
quit;
title;
In Texas, which Zip value has the highest percentage of customers? 77001
4. Splitting One Table into Many
/*s106s04.sas*/

%let table=sq.globalmetadata;
%let column=Region;

proc sql noprint;

select distinct
cat("DATA out_",
compress(&column.,,'kad'),
"; set &TABLE.(where=(&COLUMN.='", &column.,"')); run;")
length=500
into :allsteps separated by ';'
from &table.;
quit;

%macro runSteps;
&allsteps.;
%mend;

%runSteps;

How many tables were created? Eight

Solutions to Activities and Questions

continued...
6.01 Activity – Correct Answer
1. %let State=GA;
%let CreditMin=650;

proc sql;
create table Customer&State as
select CustomerID, Employed, Race,
Married, State, CreditScore The new table contains
from sq.customer
where State="&State" and rows where State equals
CreditScore > &CreditMin; GA and CreditScore is
quit; greater than 650.

NOTE: Table WORK.CUSTOMERGA created, with 957

rows and 6 columns.

continued...
6.01 Activity – Correct Answer
2. Wha t i s the na me of the newly crea ted ta ble? The table name is
CUSTOMERNC. How ma ny rows ? 700 rows
Change the values as
necessary.
%let State=NC;
%let CreditMin=700;

proc sql; The new table contains

create table Customer&State as rows where State equals
select CustomerID, Employed, Race, NC and CreditScore is
Married, State, CreditScore greater than 700.
from sq.customer
where State="&State" and
CreditScore > &CreditMin;
quit;

6.01 Activity – Correct Answer

3. How ma ny rows a re i n the new ta ble? Zero rows are returned.

%let State=NC;
%let CreditMin=700;
The macro variable does not resolve in
proc sql; single quotation marks. The WHERE
create table Customer&State as
clause is searching for the character
select CustomerID, Employed, Race,
Married, State, CreditScore string '&State'.
from sq.customer
where State='&State' and
CreditScore > &CreditMin;
quit;

continued...
6.02 Activity – Correct Answer
1.

46 %put &=MaxPop &=TotalCtry;

MAXPOP=1.1413E9 TOTALCTRY=151

When storing a number, SAS uses the BEST8. format. If the

number is larger than eight digits, scientific notation is often used.

continued...
6.02 Activity – Correct Answer
2. select distinct CountryCode, ShortName, Region,
EstYear3Pop format=comma16.
from sq.globalfull
where EstYear3Pop = &MaxPop;

1141324637 = 1.1413E9

The scientific notation value is different from

the original and thus might not find a match.

NOTE: No rows were selected.

continued...
6.02 Activity – Correct Answer
3.
proc sql noprint;
select max(EstYear3Pop) format=10.,
count(distinct CountryCode)
into :MaxPop trimmed, :TotalCtry trimmed
from sq.globalfull;
quit; The The
FORMAT
FORMATcolumn
columnmodifier
modifier
will
%put &=MaxPop &=TotalCtry; formats
format the
thestored
storedvalue
value of
ofthe
the
macro
macrovariable.
variable

46 %put &=MaxPop &=TotalCtry;

MAXPOP=1141324637 TOTALCTRY=151

6.02 Activity – Correct Answer

3. Whi ch country ha s the l argest three -year esti mated population? China
title "Country with the Largest 3 Year Estimated Population";
title2 "Out of &TotalCtry Countries";
proc sql;
select distinct CountryCode, ShortName, Region,
EstYear3Pop format=comma16.
from sq.globalfull
where EstYear3Pop = &MaxPop;
quit;
title;

Copyright © 2019, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.
Lesson 7 Accessing DBMS Data
with SAS/ACCESS®
7.1 Overview of SAS/ACCESS Technology ........................................................................ 7-3

7.2 SQL Pass-Through Facility .......................................................................................... 7-6

Demonstration: Using an SQL Pass-Through Query .................................................. 7-13

7.3 SAS/ACCESS LIBNAME Statement ............................................................................ 7-18

Demonstration: Using the SAS/ACCESS LIBNAME Statement .................................... 7-20

7.4 FEDSQL Procedure.................................................................................................... 7-27

7.5 Solutions ................................................................................................................... 7-39

Solutions to Activities and Questions........................................................................ 7-39
7-2 Lesson 7 Accessing DBMS Data w ith SAS/ACCESS®

7.1 Overview of SAS/ACCESS Technology

SAS/ACCESS Technology

SAS provides data access to more of your data sources so that you can make better decisions
faster. These interfaces are out-of-the-box solutions that provide enterprise data access and
integration between SAS and third-party databases. SAS/ACCESS interfaces enable your SAS
solutions to read, write, and update data no matter what native databases or platforms you use.

SAS/ACCESS Technology

SAS Client SAS/ACCESS Interface Engine DBMS

The SAS/ACCESS interface engine is a tool that enables you to transfer

data between the database management system (DBMS) and SAS.

For example, if you need to read data in Teradata, there is SAS/ACCESS Interface to Teradata. If
you need to read Oracle, there is SAS/ACCESS Interface to Oracle.

Considerations when Working with DBMS

For improved efficiency, process as much as possible

There are many factors affecting the optimization of your programs when working with DBMSs. The
best advice is to read the SAS/ACCESS documentation for your specific DBMS. Then try multiple
solutions to a program and benchmark the programs to see which program is running more
efficiently.

Connecting to DBMS

SQL Pass-Through LIBNAME Statement

You must have the
correct SAS/ACCESS
software installed
for your DBMS!

Code in the native DBMS Enables SAS to translate SAS

syntax SQL to native DBMS SQL

SQL pass-through is known as explicit pass-through. The LIBNAME statement is known as implicit
pass-through.
The PRODUCT_STATUS procedure returns a list of the SAS Foundation products that are installed
on your system, along with the version numbers of those products. It provides a quick method to
determine whether a SAS product is available for your use. The results from PROC
PRODUCT_STATUS are returned to the SAS log.

PROC PRODUCT_STATUS;
RUN;

For more information about the PRODUCT_STATUS procedure, see "PRODUCT_STATUS

Procedure" in the SAS® 9.4 Procedures Guide, Seventh Edition documentation. You can also find
the direct link in the Course Links section on the ELP.

7.2 SQL Pass-Through Facility

Scenario

DATA Step

PROC Step
Legacy Oracle SQL
Programs

Scenario

Top 10 customer incomes

Send native Oracle SQL from
SAS to the DBMS for
processing.

The customer data resides in your company's Oracle database, and you are already familiar with
native Oracle SQL. You would like to take pass native Oracle SQL from SAS to the database for
processing.

SQL Pass-Through Facility

Process native
Native DBMS syntax
database SQL

The DBMS processes the syntax as

if you are coding in the DBMS.

The SQL pass-through facility enables you to send DBMS-specific SQL statements directly to a
DBMS for execution. The pass-through facility uses a SAS/ACCESS interface engine to connect to
the DBMS. Therefore, you must have SAS/ACCESS software installed for your DBMS.

SQL Pass-Through Facility

1 2 3
Establish a connection Use native DBMS syntax to Terminate the
with the DBMS retrieve data to be used in a connection to the DBMS
PROC SQL query

SQL Pass-Through Facility

Establishes a connection
1
with the DBMS

PROC SQL;
CONNECT TO DBMS-name <AS alias> (DBMS-connection-options);
SELECT column-list
2
Retrieves and uses DBMS
FROM CONNECTION TO DBMS-name|alias data in a PROC SQL query
(DBMS-query);
DISCONNECT FROM DBMS-name|alias;
Native DBMS query
QUIT;

Disconnects from the

SQL Pass-Through: CONNECT

proc sql;
connect to oracle(user=sas_user pw=sastest
path=localhost);
select UserID, Income format=dollar16., State
from connection to oracle
(select UserID, Income, State
from customer
where Income is not null Make your
order by Income desc
fetch first 10 rows only); connection to Oracle
disconnect from oracle; using your specific
quit; DBMS options.

Here is what the CONNECT statement does:

• establishes the connection to the DBMS
• specifies the DBMS name
• provides connection information specific to the DBMS
• specifies an alias for the connection (optional)

Option Specifies
USER= Oracle user name
PW= Oracle password associated with the Oracle user name
PATH= Oracle driver, node, and database, or a database alias
For specific DBMS connection information, view the SAS documentation. For specific values of
connectivity options or table names, contact your DBMS administrator.

SQL Pass-Through: SAS SELECT

proc sql;
connect to oracle(user=sas_user pw=sastest
path=localhost);
select UserID, Income format=dollar16., State
from connection to oracle
(select UserID, Income, State
from customer
where Income is not null The SELECT
order by Income desc statement in PROC
fetch first 10 rows only); SQL returns results
disconnect from oracle; from the native
quit;
Oracle query.

The SELECT statement in PROC SQL consists of the following:

• a SAS SQL SELECT expression
• a FROM CONNECTION TO component
• SQL code to be passed to the DBMS
You can use SAS features such as functions, aliases, formats, and labels to customize the
pass-through query results.

SQL Pass-Through: DBMS Query

proc sql;
connect to oracle(user=sas_user pw=sastest
path=localhost);
select UserID, Income format=dollar16., State
from connection to oracle
(select UserID, Income, State
from customer
where Income is not null
order by Income desc Native Oracle SQL
fetch first 10 rows only); syntax is sent to the
disconnect from oracle; DBMS.
quit;

The query is processed by Oracle.

This query selects the Oracle DBMS customer table and is executed directly in Oracle. The DBMS
processes the syntax as if you are coding directly in the DBMS. The DBMS results are read by SAS
and processed based on the SELECT statement. You cannot use features specific to SAS inside the
parentheses. Everything inside the parenthesis is SQL that is native to the Oracle DBMS.
When using a different schema from the default, the database table in the pass -through query is
specified with the syntax schema.table_name.
The FETCH FIRST n ROWS ONLY syntax was used in Oracle Database Express Edition Release
18.4.0.0.0 (18c).

SQL Pass-Through: DISCONNECT

proc sql;
connect to oracle(user=sas_user pw=sastest
path=localhost);
select UserID, Income format=dollar16., State
from connection to oracle
(select UserID, Income, State
from customer
where Income is not null
order by Income desc It’s good practice to
fetch first 10 rows only); disconnect from the
disconnect from oracle; DBMS.
quit;

You can close the connection to the DBMS by using one of the following methods:
• submitting a DISCONNECT statement
• terminating the SQL procedure (for example, with a QUIT statement)

SQL Pass-Through
proc sql;
1 connect to oracle(user=sas_user pw=sastest
path=localhost);
2 select UserID, Income format=dollar16., State
from connection to oracle
(select UserID, Income, State Native DBMS query
from customer
where Income is not null
order by Income desc
fetch first 10 rows only);
3 disconnect from oracle;
quit;

Saving SQL Pass-Through Query Results

CREATE TABLE | VIEW name AS …;
proc sql;
connect to oracle(user=sas_user pw=sastest
path=localhost);
create table work.TotalCustomer as
select UserID, Income format=dollar16., State
from connection to oracle
…query…

table view

Query results from the SQL pass-through must be saved to a SAS table or view to be used in
subsequent SAS processes.
The CREATE VIEW statement stores the pass-through query but does not check the validity of the
query. It is important to execute the view after it has been created.

Using an SQL Pass-Through Query

Scenario
Use an SQL pass-through query to a Microsoft Access database using the SAS/ACCESS Interface
to PC Files engine.

Files
• s107d01.sas
• customer – a Microsoft Access table that contains customer information

Syntax

PROC SQL;
CONNECT TO DBMS-name <AS alias> (DBMS-connection-options);
SELECT col-name, col-name
FROM CONNECTION TO DBMS-name|alias
(DBMS-query);
DISCONNECT FROM DBMS-name|alias;
QUIT;

Notes
• The CONNECT statement establishes the connection to the DBMS.
• The SELECT statement selects the results of the DBMS query.
• The DISCONNECT statement closes the connection to the DBMS.

Demo
1. Open the s107d01.sas program in the demos folder and find the Demo section. If you have
not already done so, run the libname.sas program to define the PATH macro variable.
2. Run the Microsoft Access query. Discuss the syntax error.
3. Add a CONNECT TO PCFILES statement above the SELECT statement. After PCFILES,
add the PATH= and DBPASSWORD=SASTEST options in parentheses. End the statement
with a semicolon.
proc sql;
connect to pcfiles(path="&path/database/SQL_DB.accdb"
dbpassword=sastest);
select top 10, UserID, Income, State
from customer
order by Income desc;
quit;

4. Add the SELECT statement after the CONNECT TO statement, and select all columns using an
asterisk. Add the FROM CONNECTION TO component and reference the pcfiles database.
Enclose the original Microsoft Access query in parentheses.
proc sql;
connect to pcfiles(path="&path/database/SQL_DB.accdb"
dbpassword=sastest);
select * from connection to pcfiles
(select top 10, UserID, Income, State
from customer
order by Income desc);
quit;
5. Add the DISCONNECT FROM statement to disconnect from the pcfiles database.
proc sql;
connect to pcfiles(path="&path/database/SQL_DB.accdb"
dbpassword=sastest);
select * from connection to pcfiles
(select top 10, UserID, Income, State
from customer
order by Income desc);
disconnect from pcfiles;
quit;
6. Run the SQL pass-through query and view the results.

7. In the SELECT statement, remove the asterisk and add the columns UserID, Income, and
State. Format the Income column using the DOLLAR16. format. Run the query and view the
results.
proc sql;
connect to pcfiles(path="&path/database/SQL_DB.accdb"
dbpassword=sastest);
select UserID, Income format=dollar16., State
from connection to pcfiles
(select top 10, UserID, Income, State
from customer
order by Income desc);
disconnect from pcfiles;
quit;

7.01 Activity
Open s107a01.sas from the activities fol der a nd perform the following tasks
to pa s s Mi crosoft Acces s SQL to the da tabase:
1. Exa mi ne the native DBMS query. Run the entire query. Di d i t produce a n
error?
2. In the WHERE cl a us e, replace IS MISSING wi th IS NULL. Run the entire
query. Di d i t run s uccessfully?

SQL Pass-Through Facility Advantages

DBMS can Use DBMS-specific Use native Combine SAS features

optimize queries. functions and stored DBMS code. and DBMS-specific
procedures. features.

Here are some advantages of using the SQL pass-through facility:

• It enables the DBMS to optimize your queries. When sending your DBMS-specific query, all
processing occurs in the database and enables the database to optimize that query.
• It enables you to use specific functions and stored procedures that are DBMS specific when SAS
does not have the function that you need, or use a procedure that is already stored on the DBMS.
• If you are familiar with the DBMS syntax, you can use the native syntax inside SAS.
• It enables you to use both the power of SAS and the DBMS and combine the features of both.

SQL Pass-Through Facility Considerations

Query results must be saved DBMS-specific Only SQL code within

to use in subsequent SAS SQL knowledge is the parentheses is
processes. required. passed to the DBMS.

Here are some SQL considerations related to the pass-through facility:

• To use the query results from an SQL pass-through in other SAS processes, you must save the
results as a SAS table or view.
• You must have DBMS-specific SQL knowledge to use the pass-through facility.
• Only the code within the parentheses is passed to the DBMS. WHERE processing and SORT
requests used in procedures that reference a PROC SQL pass -through view must be performed
by SAS.

Syntax Summary

PROC SQL;
CONNECT TO DBMS-name <AS alias> (DBMS-connection-options);
SELECT column-list
FROM CONNECTION TO DBMS-name|alias
(DBMS-query);
DISCONNECT FROM DBMS-name|alias;
QUIT;
SQL Pass-Through
24
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

7.3 SAS/ACCESS LIBNAME Statement

SAS/ACCESS LIBNAME Engine

SAS Engine

PROC SQL syntax Translate PROC Process native

SQL to native database SQL
DBMS SQL when
possible

The SAS/ACCESS LIBNAME engine translates your PROC SQL syntax to native DBMS SQL when
possible. This is important because SAS SQL and native DBMS SQL can differ. So if you are
working with multiple DBMSs, you can learn SAS SQL instead of learning native DBMS SQL
implementations, and let the engine do the work for you.

SAS/ACCESS LIBNAME Statement

LIBNAME libref engine <SAS/ACCESS-engine-opti ons> ;

name of name of DBMS connection

library SAS/ACCESS engine options

DBMS connection options are

dependent on the SAS/ACCESS
engine that you are using.

SAS library

The SAS/ACCESS LIBNAME statement

• uses the SAS/ACCESS engine to assign a libref to a DBMS
• enables you to reference a DBMS object directly in a DATA step or SAS procedure
• has many options to control the connection to the DBMS.

Closing the DBMS Connection

LIBNAME libref CLEAR;

Release the DBMS and It’s good practice to

associated resources. clear the DBMS
connection when
complete.

Submit a LIBNAME statement with a CLEAR option to release the DBMS and associated resources.

Using the SAS/ACCESS LIBNAME Statement

Scenario
Use the SAS/ACCESS LIBNAME statement to read a table in Microsoft Access.

Files
• s107d02.sas
• customer – a Microsoft Access table that contains customer information

Syntax

LIBNAME libref engine <SAS/ACCESS-engine-options>;

LIBNAME libref CLEAR;

Notes
The SAS/ACCESS LIBNAME statement
• uses the SAS/ACCESS engine to assign a libref to a DBMS
• enables you to reference a DBMS object directly in a DATA step or SAS procedure
• has many options to control the connection to the DBMS.

Demo
1. Open the s107d02.sas program in the demos folder and find the Demo section. If you have not
already done so, run the libname.sas program to define the PATH macro variable.
2. Begin by viewing all available libraries in your current session.
Note: Your libraries might differ. Notice that there is no library named db.

3. Complete the LIBNAME statement to define a library named db that uses the PCFILES engine.
Add the PATH= option to connect to the SQL_DB.accdb Microsoft Access database, and add
the DBPASSWORD= option using the password sastest. Highlight the LIBNAME statement and
run the selected code. Use the navigation pane to expand the db library.
Note: In SAS Enterprise Guide, click Libraries and select Refresh to update the library list.
libname db pcfiles path="&path/database/SQL_DB.accdb"
dbpassword=sastest;

4. Review the query below the LIBNAME statement. Add the library db to the beginning of the
customer table name, and apply the DOLLAR16. format to the Income column.
libname db pcfiles path="&path/database/SQL_DB.accdb"
dbpassword=sastest;

proc sql outobs=10;

select UserID, Income format=dollar16., State
from db.customer
order by Income desc;
quit;
5. Add a statement to clear the db library. Highlight the entire demo program and run the selected
code.
libname db pcfiles path="&path/database/SQL_DB.accdb"
dbpassword=sastest;

proc sql outobs=10;

select UserID, Income format=dollar16., State
from db.customer
order by Income desc;
quit;

libname db clear;

SAS/ACCESS LIBNAME Statement Advantages

Fewer lines of DBMS-specific You have The LIBNAME

SAS code are SQL knowledge is direct access engine optimizes
required. not required. to DBMS data. processing.

The LIBNAME statement offers the following advantages:

• Significantly fewer lines of SAS code are required to perform operations in your DBMS. For
example, a single LIBNAME statement establishes a connection to your DBMS, enables you to
specify how your data is processed, and enables you to easily browse your DBMS tables in SAS.
• You do not need to know your DBMS's SQL language to access and manipulate your DBMS data.
• Direct and transparent access to DBMS data. You can use SAS procedures, such as PROC SQL,
or DATA step programming on any libref that references DBMS data. You can read, insert, update,
delete, and append data, as well as create and drop DBMS tables by using normal SAS syntax.
The engine can also translate some SAS processes into the DBMS for processing, which reduces
the data that needs to be returned to SAS.
• The LIBNAME engine optimizes the processing of joins and WHERE clauses by passing these
operations directly to the DBMS to take advantage of the indexing and other processing
capabilities of your DBMS.

SAS/ACCESS LIBNAME Statement Considerations

Do what it can in the DBMS

SAS Engine

Behind the scenes, the SAS/ACCESS engine attempts to convert your SAS code to an SQL query
that is passed to the DBMS. This enables as much work to be done in the DBMS as possible.

SAS/ACCESS LIBNAME Statement Considerations

SAS Engine

For operations that cannot be converted to DBMS SQL, the SAS/ACCESS

engine generates a DBMS query to bring the data into SAS.
32
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

If SAS fails to pass the query (or parts of the query) to the DBMS, then SAS processes the code.
Things like DATA step merge, specific functions, specific data set options, or combining different
data sources must be done in SAS. The data is transferred between the two.

SAS/ACCESS LIBNAME Statement Considerations

SAS Engine
For large DBMS tables, this can result in slow performance,
because large data volume might need to be returned to SAS.

In order for PROC SQL code to be converted to SQL, the code must meet these basic requirements:
• refer to a single SAS/ACCESS LIBNAME engine
• use SAS functions that can be translated into DBMS functions
See the documentation for your specific DBMS for more information.
To examine the SQL that SAS/ACCESS submits, use the following option:

OPTIONS SASTRACE=',,,d' SASTRACELOC=saslog NOSTSUFFIX;

The SASTRACE= option enables you to examine the SQL that the SAS/ACCESS engine submits
to the DBMS.
• ',,,d' – Specifies that all SQL statements sent to the DBMS are sent to the log.
• ',,,s' – Specifies that a summary of timing information for calls made to the DBMS is sent
to the log.
The SASTRACELOC= option defines where to send the SASTRACE information.
The NOSTSUFFIX option limits the amount of information displayed in the log.
For more information about how to determine where the processing is occurring, visit the SAS
documentation or the ELP.

Working with DBMS

SAS connects Functions are not Combining different

differently to DBMSs. consistent across data sources must be
DBMSs. done in SAS.

Visit the SAS documentation.

For more information about DBMS-specific references, see the SAS/ACCESS ® 9.4 for Relational
Databases: Reference, Ninth Edition documentation. You can also find the direct link in the Course
Links section on the ELP.

7.02 Activity
Open a brows er a nd perform the following tasks to vi ew SAS DBMS -Specific
Reference documentation:
1. In a web brows er, a ccess SAS Help a t
http://s upport.s as.com/documentation.
2. Under Popular Documentation, s el ect Programming: SAS 9.4 and Viya.
3. Scrol l down until you fi nd the Accessing Data s ection. Inside the s ection,
fi nd SAS/ACCESS a nd s elect Relational Databases.
4. Scrol l down a nd find the DBMS-Specific Reference s ection a nd s elect a
reference of your choi ce.
5. Revi ew the topics i n your DBMS-s pecific reference. Cl ick the Passing SAS
Functions to l i nk a nd review the SAS functions tha t pass to your DBMS
for proces s ing.
35
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

Syntax Summary

LIBNAME libref engine <SAS/ACCESS-engine-options> ;

LIBNAME Statement

LIBNAME libref CLEAR;

Clear the libref

7.4 FEDSQL Procedure

What Is PROC FEDSQL?

PROC
FEDSQL

Vendor Neutral
ANSI SQL

Vendor neutral ANSI standard SQL

allows for more processing to be
pushed into a DBMS for processing.

PROC FEDSQL is not a replacement for PROC SQL. On the contrary, both are tools that are used
for specific scenarios. They offer different strengths for different situations. But in the end, they are
both an implementation of SQL and have some slight differences in syntax . But when you know one,
you can easily transition to another.

PROC FEDSQL Advantages: ANSI 3

ANSI 3 Compliant
First major revision Last major revision
to the ANSI to the ANSI
Standard (SQL-2) Standard (SQL-3)

PROC SQL PROC

FEDSQL

1992 1993 1999 2013

The FEDSQL procedure is a SAS proprietary implementation of ANSI SQL:1999 (SQL-3) core
standard. It provides ANSI 1999 core compliance features and proprietary extensions.
For more information about PROC FEDSQL and the ANSI Standard, see "FedSQL and the ANSI
Standard" in the SAS® 9.4 FedSQL Language Reference, Fifth Edition documentation. You can also
find the direct link in the Course Links section on the ELP.

PROC FEDSQL Advantages: Data Types

PROC SQL PROC FEDSQL

Limited to SAS data types Processes 17 ANSI data types

multiple data types

PROC FEDSQL supports many more data types than previous SAS SQL implementations.
Traditional DBMS access through SAS/ACCESS LIBNAME engines translates target DBMS data
types to and from two legacy SAS data types: SAS numeric and SAS character. PROC FEDSQL
processes ANSI data types at native precision using a threaded driver instead of the typical data
access and translation by the LIBNAME engine.

Large Data Types

proc sql; proc fedsql;

select * select *
from orc.bigint; from orc.bigint;
quit; quit;

PROC SQL loses

PROC FEDSQL allows
precision after 15
for larger data types.
significant digits.
41
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

PROC SQL
ANSI Type Resulting SAS Type Default Length
CHAR(n) Character 8
VARCHAR(n) Character 8
INTEGER Numeric 8
Traditional DBMS access translates target
SMALLINT Numeric 8 DBMS data types to SAS data types.
DECIMAL Numeric 8
NUMERIC Numeric 8
FLOAT Numeric 8
REAL Numeric 8
DOUBLE PRECISION Numeric 8
DATE Numeric 8

PROC SQL can work only with the data types defined in SAS: numeric and character. When working
with DBMSs, SAS translates ANSI data types to SAS data types.
For example, when working with a DBMS that has data defined as integer or float , that is not a
problem for PROC SQL it converts the value to a SAS numeric. The precision values of SAS
numeric values are accurate to approximately 15 digits.

For more information about data types for SAS data, see "Data Types for SAS Data Sets" in the
SAS® 9.4 FedSQL Language Reference, Fifth Edition documentation. You can also find the direct
link in the Course Links section on the ELP.
For more information about the numerical accuracy in SAS software, see "Numerical Accuracy in
SAS Software" in the SAS® 9.4 FedSQL Language Reference, Fifth Edition documentation. You can
also find the direct link in the Course Links section on the ELP.

PROC FEDSQL Data Types

DATA TYPES
BIGINT FLOAT(p) TIME(p)
BINARY(n) INTEGER TIMESTAMP(p)
CHAR(n) NCHAR(n) TINYINT
DATE NVARCHAR(n) VARBINARY(n)
DOUBLE REAL VARCHAR(n)
DECIMAL|NUMERIC(p,s) SMALLINT

PROC FEDSQL allows for 17 ANSI data types.

For more information about PROC FEDSQL data types, see "Data Types" in the SAS® 9.4 FedSQL
Language Reference, Fifth Edition documentation. You can also find the direct link in the Course
Links section on the ELP.
For more information about PROC FEDSQL data types for specific DBMSs, see "Data Type
Reference" in the SAS® 9.4 FedSQL Language Reference, Fifth Edition documentation. You can
also find the direct link in the Course Links section on the ELP.

PROC FEDSQL Advantages: Multi-Threading

PROC SQL PROC FEDSQL

Is multi-threaded for sorting and Is fully multi-threaded on the SAS
indexing on the SAS Platform Platform

Sort/index Everything else

PROC FEDSQL provides a scalable, threaded, high-performance way to access, manage, and
share relational data in multiple data sources. When possible, PROC FEDSQL queries are optimized
with multi-threaded algorithms in order to resolve large-scale operations.

PROC FEDSQL DICTIONARY Tables

FedSQL DICTIONARY
COLUMNS tables provide an
CATALOGS in-depth look into your
COLUMN_ database tables and
STATISTICS columns.

TABLES

STATISTICS

A PROC FEDSQL DICTIONARY table is a Read-only table that contains information about columns,
tables, and catalogs, as well as statistics about tables and their associated indexes.
PROC FEDSQL DICTIONARY tables can give you a more in-depth look into your database tables
and columns.

For more information about PROC FEDSQL DICTIONARY tables, see "FedSQL DICTIONARY
Tables" in the SAS® 9.4 FedSQL Language Reference, Fifth Edition documentation. You can also
find the direct link in the Course Links section on the ELP.

PROC FEDSQL Syntax

PROC FEDSQL <options>;

SELECT col-name, col-name The foundation of

FROM input-table the PROC FedSQL
syntax is similar to
<WHERE clause>
PROC SQL.
<GROUP BY clause>
<HAVING clause>
<ORDER BY clause>;

QUIT;

When using PROC FEDSQL, ANSI standard syntax is required.

PROC FEDSQL LIBNAME Statement

LIBNAME libref engine <SAS/ACCESS-engine-options>;

PROC FEDSQL <options>;

SELECT col-name, col-name
PROC FEDSQL needs the
FROM libref.table; correct DBMS information
QUIT; through the SAS LIBNAME
statement.

PROC FEDSQL LIBNAME Statement

libname mkt oracle user=sas_user password=sastest

path=localhost;

proc fedsql;
select State,
count(*) as TotalCustomer
from mkt.customer
where CreditScore > 700 Reference the table.
group by State
order by TotalCustomer desc;
quit;

libname mkt clear;

7.03 Activity
Open s107a03.sas from the activities fol der a nd perform the following tasks
to perform a PROC FEDSQL query:
1. Exa mi ne a nd run the query. Di d i t produce a n error?
2. In the WHERE cl a us e, replace the double quotation marks a round NC
wi th single quotation marks. Run the query. Di d i t run s uccessfully?

Examining the SQL That PROC FEDSQL Submits

PROC FEDSQL IPTRACE;

libname mkt oracle user=sas_user

The IPTRACEpassword=sastest
option prints a text description
path=localhost;
of the PROC FEDSQL query plan.
proc fedsql iptrace;
select State,
count(*) as TotalCustomer
from mkt.customer
where CreditScore > 700
group by State
order by TotalCustomer desc;
quit;

libname mkt clear;

For debugging purposes, you can see that PROC FEDSQL submits to the DBMS by specifying the
undocumented IPTRACE option. This option causes PROC FEDSQL to write the SQL sent to the
RDBMS in the log as the "Retextualized child query." If the query is not accepted by the DBMS,
IPTRACE writes the error message generated by the RDBMS in the log. PROC FEDSQL continues
to reformulate the query and try again until the DBMS executes the query and returns a result set.
IPTRACE then reports how much of the original query was pushed down into the DBMS.
Another method that prints the query plan is the _METHOD option. For more information about
_METHOD, see "_METHOD FedSQL Option" in the Base SAS ® 9.4 Procedures Guide, Seventh
Edition documentation. You can also find the direct link in the Course Links section on the E LP.

Working with SAS Viya

PROC SQL PROC FEDSQL

Does not execute in CAS Executes in CAS

SAS Viya on the SAS Platform

SAS Platform

SAS Viya
• an open, cloud-enabled, analytic run-time environment

SAS Cloud Analytic Services (CAS)

in-memory fast data of

engine processing any size

SAS Viya is the latest enhancement of the SAS Platform. It is an open, cloud-enabled, analytic run-
time environment with a number of supporting services.
One of those supporting services is SAS Cloud Analytic Services, or CAS. CAS provides a powerful
in-memory engine that delivers blazing speed to accurately process your big data. It uses scalable,
high-performance, multi-threaded algorithms to rapidly perform analytical processing on in-memory
data of any size.

PROC FEDSQL Disadvantages

PROC SQL PROC FEDSQL
Includes many non-ANSI standard Includes very few non-ANSI SAS
SAS enhancements enhancements

Mnemonics

Limited
Column INTO
Modifiers

Limited SAS
Functions

Limited
Options Calculated

PROC SQL PROC FEDSQL

SAS SQL implementation Vendor neutral ANSI SQL

Follows ANSI Standard 2 ANSI Standard 3 Compliant

Limited to SAS data types Processes 17 ANSI data types

Is multi-threaded for sorting and Is fully multi-threaded on the SAS

indexing on the SAS Platform Platform

Includes many non-ANSI standard SAS Includes very few non-ANSI SAS
enhancements enhancements

Does not execute in CAS Executes in CAS

For a summary document about PROC SQL versus PROC FEDSQL, see "PROC SQL vs. PROC
FEDSQL Summary" on the ELP.

Syntax Summary

PROC FEDSQL <options>;

SELECT col-name, col-name
FROM input-table
<WHERE clause> PROC FEDSQL IPTRACE;
<GROUP BY clause> PROC FEDSQL Options
<HAVING clause>
<ORDER BY clause>;
QUIT;
PROC FEDSQL

Beyond SQL Programming

What if you want to ...

. . . l ea rn more a bout worki ng wi th DBMSs by . . . s ee the ELP for

vi ewi ng the SAS documentation or ta king a a ddi tional i nforma tion
SAS cours e a bout worki ng wi th DBMS
• View the SAS/ACCESS 9.4 for Relational Databases: • View the Working with
Reference, Ninth Edition. Database section in the ELP for
• View the SAS 9.4 FedSQL Language Reference, Fifth Edition. additional SAS papers and
• Take the Processing Database and PC File Data with blogs.
SAS/ACCESS Software course.
• Take the Introduction to SAS/ACCESS Interface to Teradata
course.
• Take the Introduction to SAS and Hadoop course.

Put It All Together!

Practice all your new SAS

skills by completing a
comprehensive case study.
Visit the Extended Learning
page to access the case study
materials.

Case Study

Join the Discussion

https://communities.sas.com/sas-training

Visit the SAS Training

Community to ask
questions of our experts
and exchange ideas with
other SAS users.

7.5 Solutions
Solutions to Activities and Questions

7.01 Activity – Correct Answer

1. Di d i t produce a n error? Yes
2. Di d i t run s uccessfully? Yes. IS MISSING is a SAS enhancement. IS NULL
is ANSI standard.

connect to pcfiles(path="&path\database\SQL_DB.accdb"
dbpassword=sastest);
select UserID, Income format=dollar16., State
from connection to pcfiles
(select top 10, UserID, Income, State
from customer
where BankID is null A native ACCESS SQL
order by Income desc); query requires IS NULL.
disconnect from pcfiles;

7.03 Activity – Correct Answer

1. Di d i t produce a n error? Yes
2. Di d i t run s uccessfully? Yes. Double quotation marks around a character
string is a SAS enhancement, and it produces an error when using PROC
FEDSQL. Single quotation marks around a character string is ANSI
standard and required in FEDSQL.

proc fedsql;
select UserID, Income, State Single quotation marks
from sq.customer are ANSI standard.
where State='NC'
order by Income desc
The LIMIT statement limits the
limit 10;
quit; number of records returned in
FedSQL.
50
C o p yri gh t © SAS In sti tu te In c. Al l ri gh ts re se rve d .

To use a macro variable in a character string in PROC FEDSQL, you must use
%TSLIT(&macro-variable).

%LET StateVar=NC;

PROC FEDSQL;
SELECT UserID, Income, State
FROM sq.customer
WHERE State=%tslit(&StateVar)
ORDER BY Income DESC
LIMIT 10;
QUIT;

ReinventingtheOrganization A5
No ratings yet
ReinventingtheOrganization A5
6 pages
Sas 1
100% (1)
Sas 1
292 pages
SAS® Programming 3 Advanced Techniques PDF
100% (2)
SAS® Programming 3 Advanced Techniques PDF
244 pages
SAS DI Developer - Fast Track PDF
100% (8)
SAS DI Developer - Fast Track PDF
982 pages
Blockchain and Distributed Ledger Technologies
No ratings yet
Blockchain and Distributed Ledger Technologies
34 pages
PG1
63% (8)
PG1
294 pages
SAS Enterprise Guide Querying and Reporting - Course Notes PDF
100% (1)
SAS Enterprise Guide Querying and Reporting - Course Notes PDF
292 pages
SAS Programming 2 Data Manipulation Techniques - Quizzes PDF
100% (1)
SAS Programming 2 Data Manipulation Techniques - Quizzes PDF
92 pages
SAS Programming II Data Manipulation Techniques
No ratings yet
SAS Programming II Data Manipulation Techniques
2 pages
Data Preparation For Analytics Using SAS
100% (1)
Data Preparation For Analytics Using SAS
440 pages
Sas Course Content Porak Technologies
No ratings yet
Sas Course Content Porak Technologies
8 pages
SAS DI Developer - Fast Track PDF
No ratings yet
SAS DI Developer - Fast Track PDF
982 pages
Creating Business Intelligence For Your Organization Fast Track
0% (1)
Creating Business Intelligence For Your Organization Fast Track
708 pages
A Complete Tutorial On SAS Macros For Faster Data Manipulation
No ratings yet
A Complete Tutorial On SAS Macros For Faster Data Manipulation
36 pages
Reading Comprehension
100% (6)
Reading Comprehension
2 pages
Listado Escalas
100% (1)
Listado Escalas
108 pages
SQL Processing With SAS
No ratings yet
SQL Processing With SAS
232 pages
Ten Good Reasons To Learn Sas Software'S SQL Procedure: Sigurd W. Hermansen, Westat, Rockville, MD
No ratings yet
Ten Good Reasons To Learn Sas Software'S SQL Procedure: Sigurd W. Hermansen, Westat, Rockville, MD
5 pages
LWEG171 E70539 05apr2016p374 v11
No ratings yet
LWEG171 E70539 05apr2016p374 v11
374 pages
SAS Macro Language Reference PDF
No ratings yet
SAS Macro Language Reference PDF
385 pages
The Essential PROC SQL
No ratings yet
The Essential PROC SQL
10 pages
LWMC1V2 001
100% (2)
LWMC1V2 001
296 pages
Sas Data Management
100% (1)
Sas Data Management
908 pages
Merging Data Seven Different Ways
100% (1)
Merging Data Seven Different Ways
1 page
Proc SQL
100% (1)
Proc SQL
7 pages
Base Sas Programming 2019 10 04 13 47 36 122
No ratings yet
Base Sas Programming 2019 10 04 13 47 36 122
281 pages
Guide To SAS 9.4 & SAS Viya 3.3 Programming Documentation
100% (1)
Guide To SAS 9.4 & SAS Viya 3.3 Programming Documentation
4 pages
Base SAS 9.4 Procedures Guide High-Performance Procedures, Third Edition
No ratings yet
Base SAS 9.4 Procedures Guide High-Performance Procedures, Third Edition
172 pages
SAS Macro That Reads The Filenames Available at A Particular Directory On Any FTP Server
No ratings yet
SAS Macro That Reads The Filenames Available at A Particular Directory On Any FTP Server
9 pages
SAS 9.4 Graph Template Language Reference
No ratings yet
SAS 9.4 Graph Template Language Reference
1,454 pages
SAS Exam
100% (1)
SAS Exam
32 pages
Exploring Sas Viya
No ratings yet
Exploring Sas Viya
78 pages
Base SAS 9.4 Procedures Guide, Seventh Edition
No ratings yet
Base SAS 9.4 Procedures Guide, Seventh Edition
2,828 pages
SAS Macro Language Reference
No ratings yet
SAS Macro Language Reference
385 pages
SAS Technical Interview Questions
No ratings yet
SAS Technical Interview Questions
4 pages
Visual Analytics With SAS Viya
No ratings yet
Visual Analytics With SAS Viya
148 pages
Sascheatsheet 170401221255
100% (1)
Sascheatsheet 170401221255
29 pages
Data Management With SAS
100% (2)
Data Management With SAS
88 pages
SAS+Programming+Fast+Track+ Cource+Notes
No ratings yet
SAS+Programming+Fast+Track+ Cource+Notes
406 pages
SAS Basics
100% (1)
SAS Basics
29 pages
SAS Do Array
100% (1)
SAS Do Array
118 pages
SAS Proc Report at Work
No ratings yet
SAS Proc Report at Work
43 pages
(Statistical Analysis System) : By: Kirtikrushna
No ratings yet
(Statistical Analysis System) : By: Kirtikrushna
129 pages
Vds
100% (1)
Vds
220 pages
Base Sas Advance SAS Interview Questions by Keli Technologies
No ratings yet
Base Sas Advance SAS Interview Questions by Keli Technologies
12 pages
SAS Interview Practice Question With Answers
100% (1)
SAS Interview Practice Question With Answers
50 pages
SAS Programming 3 Advanced Techniques and Efficiencies
100% (1)
SAS Programming 3 Advanced Techniques and Efficiencies
888 pages
Sas Macros Final
No ratings yet
Sas Macros Final
75 pages
Sas Macros
No ratings yet
Sas Macros
247 pages
SAS Macro Basics
No ratings yet
SAS Macro Basics
10 pages
Analysis Using SAS
100% (1)
Analysis Using SAS
40 pages
Basics of SAS 2
No ratings yet
Basics of SAS 2
417 pages
SAS Manipulate Datasets
No ratings yet
SAS Manipulate Datasets
32 pages
SAS Slides 7: Match Merging With Datastep
No ratings yet
SAS Slides 7: Match Merging With Datastep
22 pages
Sas fAQ'S 1
No ratings yet
Sas fAQ'S 1
114 pages
Read Text: Sas My Code
No ratings yet
Read Text: Sas My Code
6 pages
Advanced PROC SQL Programming Techniques
No ratings yet
Advanced PROC SQL Programming Techniques
6 pages
Base SAS Interview Questions
No ratings yet
Base SAS Interview Questions
26 pages
SAS Interview Questions and Answers
No ratings yet
SAS Interview Questions and Answers
107 pages
SAS Training - Day1
No ratings yet
SAS Training - Day1
36 pages
SAS Interview Questions You'll Most Likely Be Asked
From Everand
SAS Interview Questions You'll Most Likely Be Asked
Vibrant Publishers
No ratings yet
Facility Management: Business Process Integration
From Everand
Facility Management: Business Process Integration
Alexander Redlein
No ratings yet
Breakthrough Improvement with QI Macros and Excel: Finding the Invisible Low-Hanging Fruit: Finding the Invisible Low-Hanging Fruit
From Everand
Breakthrough Improvement with QI Macros and Excel: Finding the Invisible Low-Hanging Fruit: Finding the Invisible Low-Hanging Fruit
Jay Arthur
No ratings yet
Competencies: How they are acquired and measured
From Everand
Competencies: How they are acquired and measured
Dieter Gnahs
No ratings yet
State of The Nation Address of Pres. Rodrigo "Digong" Duterte
No ratings yet
State of The Nation Address of Pres. Rodrigo "Digong" Duterte
2 pages
Taylor Complaint Disbarment/Prosecution To WASHINGTON STATE BAR ASSOCIATION
No ratings yet
Taylor Complaint Disbarment/Prosecution To WASHINGTON STATE BAR ASSOCIATION
45 pages
Guidelines For Unit 1 Biology 2021-2022
No ratings yet
Guidelines For Unit 1 Biology 2021-2022
4 pages
Level Instruments: G.E.T. Training (DP Type, Magnétostrictive, GWR Type, Non Contact Type RADAR Etc.,)
No ratings yet
Level Instruments: G.E.T. Training (DP Type, Magnétostrictive, GWR Type, Non Contact Type RADAR Etc.,)
22 pages
Topeak MTX DXP
No ratings yet
Topeak MTX DXP
1 page
Ranjeet Singh
No ratings yet
Ranjeet Singh
3 pages
Reasons of Power Theft:: The Electricity Regulatory Commission Act, 1998
No ratings yet
Reasons of Power Theft:: The Electricity Regulatory Commission Act, 1998
3 pages
4th COT
No ratings yet
4th COT
4 pages
Medication Math
100% (1)
Medication Math
48 pages
Office of The Mayor: Name Form of Expression Creator Year/Era of Creation Description State
No ratings yet
Office of The Mayor: Name Form of Expression Creator Year/Era of Creation Description State
19 pages
Mathematical Statistics I Muzammil Tanveer
No ratings yet
Mathematical Statistics I Muzammil Tanveer
64 pages
Grass Family
No ratings yet
Grass Family
55 pages
Dynamic Earth Summary
No ratings yet
Dynamic Earth Summary
7 pages
Schletter-Product Sheets-Ground Mounted Systems-FS Duo
No ratings yet
Schletter-Product Sheets-Ground Mounted Systems-FS Duo
2 pages
Modern Banking Full Notes
100% (1)
Modern Banking Full Notes
49 pages
Science7 Ic WS#29
No ratings yet
Science7 Ic WS#29
2 pages
SUNDAY TIMES HERITAGE PROJECT - Buffalo City
No ratings yet
SUNDAY TIMES HERITAGE PROJECT - Buffalo City
66 pages
Resume 18 04 2024 02 42 27 PM
No ratings yet
Resume 18 04 2024 02 42 27 PM
5 pages
Business Studies Paper 2 Revision Amended
No ratings yet
Business Studies Paper 2 Revision Amended
6 pages
The Inn at Five Points
100% (9)
The Inn at Five Points
16 pages
Bronze Full Way Gate Valve BS 5154 PN20 Series B
No ratings yet
Bronze Full Way Gate Valve BS 5154 PN20 Series B
11 pages
PHOTON - AI Powered PV O&M System - Feature - Compressed
No ratings yet
PHOTON - AI Powered PV O&M System - Feature - Compressed
36 pages
Diagnosis and Treatment Planning With 3-Dimensional Dentofacial Records
No ratings yet
Diagnosis and Treatment Planning With 3-Dimensional Dentofacial Records
9 pages
Shear Force and Bendind Moment
No ratings yet
Shear Force and Bendind Moment
8 pages
Research Progress in The Development of Natural Gas As Fuel For Road Vehicles: A Bibliographic Review (1991-2016)
No ratings yet
Research Progress in The Development of Natural Gas As Fuel For Road Vehicles: A Bibliographic Review (1991-2016)
41 pages
VodafonePortugalMilestones
No ratings yet
VodafonePortugalMilestones
33 pages