OLAP

Textbook Chapter 33
March 19, 2019
Today

• OLTP vs OLAP
• Data Warehouses
• Extract, Transform, Load
• Multidimensional Data Models
• SQL:
• GROUPING SETS
• Window Functions

CMPT 355: Theory and Application of Databases 2

 Textbook Chapter 31
Data Warehouses

• A Data Warehouse is a type of database that typically geared toward

analysis of historical data.
• This analysis is called OLAP (Online Analytical Processing)
• Requires highly temporal data that might not be required in a normal
database
• A traditional database supports transaction processing
• This is called OLTP (Online Transactional Processing)
• They are not equipped for handing major analytics
• Some amount of temporal data is often required in these as well

CMPT 355: Theory and Application of Databases 3

 https://en.wikipedia.org/wiki/Online_transaction_processing

OLTP vs. OLAP https://en.wikipedia.org/wiki/Online_analytical_processing

Textbook Section 31.1

OLTP: Online transaction processing OLAP: Online analytical processing

• What we’ve been doing this entire course. • Usually querying a purpose-built “data
• Implemented with relational DBMS warehouse”
• Often a copy of the data

System Design System Design

• Fully normalized • Optimized for read performance
• Built for data entry • May or may not be normalized
• Might not support data entry
Intended Usage
• Support day-to-day activities and operational Intended Usage
processing • Analytical, read-only queries
• CRUD operations • Business intelligence or reporting
• Data mining

CMPT 355: Theory and Application of Databases 4

 Textbook Table 31.1
OLTP vs. OLAP (data warehouses)
Characteristic OLTP Systems Data Warehousing Systems
Main Purpose Support operational processing. Support analytical processing.
Data age Current, real-time data. Historic.
Sometimes current data also included.
Data latency Real-time. Updated on a fixed schedule.
Some systems might be real-time.
Data granularity Detailed data. Detailed data along with summarized data.
Data processing Predictable CRUD operations. Less predictable.
High throughput of transactions. Low to medium throughput of transactions.
Reporting Predictable, one-dimensional, relatively static Unpredictable, multidimensional, dynamic
fixed reporting. reporting.
Users Serves large number of operational users. Serves lower number of users.
Possibly supports analytical requirements of
operational users.

CMPT 355: Theory and Application of Databases 5

Why?

• In this course, we’ve worked with relatively small data sets.

• ~100 rows in Assignment 2
• One table had 11,203 rows
• <2000 rows in Assignment 3
• <4000 rows in Assignment 4
• A single user (you) was querying the DB.
• In these cases, a data warehouse is overkill.
• But what about real-world databases?

CMPT 355: Theory and Application of Databases 6

 https://data.stackexchange.com/
Why Data Warehouses?

• Consider Stack Overflow

CMPT 355: Theory and Application of Databases 7

 https://meta.stackexchange.com/
questions/250396/database-
diagram-of-stack-exchange-
model/250439

CMPT 355: Theory and Application of Databases 8

Stack Overflow DB

SELECT
Posts.Id,
Users.Id
FROM Posts
JOIN Users
ON Users.Id =
Posts.OwnerUserId

CMPT 355: Theory and Application of Databases 9

Why Create a Data Warehouse?

• You don’t want to take resources away from OLTP database.

• When extracting data from OLTP database, you can transform it in
various useful ways.
• Clean out bad or invalid data
• Perform calculations
• Standardize data formats
• You can structure the data in a “flatter” way that removes need for
complex JOINs, improves read performance, or is more accessible to
end-users of the OLAP system.
• The StackOverflow DB isn’t bad for this.
From Posts, you can get to any other entity with no more than 2 JOINs.

CMPT 355: Theory and Application of Databases 10

Why Create a Data Warehouse?

• You don’t want to take resources away from OLTP database.

• When extracting data from OLTP database, you can transform it in
various useful ways.
• Clean out bad or invalid data ETL – Extract, Transform, Load
• Perform calculations
• Standardize data formats Dimensional Modelling
• You can structure the data in a “flatter” way that removes need for
complex JOINs, improves read performance, or is more accessible to
end-users of the OLAP system
• The StackOverflow DB isn’t bad for this.
From Posts, you can get to any other entity with no more than 2 JOINs.

CMPT 355: Theory and Application of Databases 11

Extract, Transform, Load

• A process of migrating data often used in data warehousing

CMPT 355: Theory and Application of Databases 12

Extract, Transform, Load

• Data Extraction
• Take data from one (or many) source(s).
• Data Transformation
• Modify/transforming the data into the format required.
• Almost always requires cleaning of data to:
• Remove bad data/characters
• Making list of values consistent
• Standardize data formatting
• Data Load
• Load the transformed data into your new system.

CMPT 355: Theory and Application of Databases 13

Extract, Transform, Load

• This is your opportunity to clean up the

data!
• If your “province” column looks like this,
how can you look at sales per province?

CMPT 355: Theory and Application of Databases 14

ETL for Pokemon DB: Extraction

I did this for you in the following steps:

• The data was extracted from web pages
• e.g., List of Pokemon from Wikipedia:
https://en.wikipedia.org/wiki/List_of_generat
ion_I_Pok%C3%A9mon
• I copied and pasted into Excel.
• Removed all the mega-evolutions.
• Removed all the columns that we weren’t
going to import.
• Cleaned up header row

CMPT 355: Theory and Application of Databases 15

ETL for Pokemon DB: Extraction

• I split all of the merged cells for

single-type Pokemon so that
there would be a blank second
cell.
• I exported to CSV to eliminate
formatting an the links.
• And this left me with something I
could safely import into a SQLite
DB.

CMPT 355: Theory and Application of Databases 16

ETL for Pokemon DB: Transform

After importing into SQLite, we had to

transform the data.
• Find the numeric TypeID from named type
• Switch from “EvolvesTo” to “EvolvesFrom”

This is also where we had to clean up

the data.
• NULLs instead of blank strings

CMPT 355: Theory and Application of Databases 17

ETL for Pokemon DB: Transform

For other tables, we had to do

some extra cleaning up.
• Remove unneeded whitespace
• Make ‘?’ and ‘-’ NULL

I had you do all this in SQL, but you can

do it outside the DB as well.

CMPT 355: Theory and Application of Databases 18

ETL for Pokemon DB: Load

INSERT INTO Pokemon (

• And then finally, to load, we PokedexNumber, Name, PrimaryTypeID,
had to write an INSERT INTO … SecondaryTypeID, EvolvesFromNumber
)
SELECT … query. SELECT
PT1.PokemonNumber,
PT1.Name,
T1.TypeID AS PrimaryTypeID,
T2.TypeID AS SecondaryTypeID,
PT1.EvolvesTo,
PT2.PokemonNumber AS EvolvesFrom
FROM pokemon_temp AS PT1
JOIN Types AS T1
ON T1.Name = TRIM(PT1.PrimaryType)
LEFT JOIN Types AS T2
ON T2.Name = TRIM(PT1.SecondaryType)
LEFT JOIN pokemon_temp PT2
ON PT2.EvolvesTo = PT1.PokemonNumber
CMPT 355: Theory and Application of Databases 19
ETL for Assignment 4

• For Assignment 4, I had you take unnormalized, extracted, data, and

then Load it based on your design and the 3NF version of the data.
• To Extract that data, I
• Created my own DB schema for a hypothetical “old” database.
• Generated mock data.
• Wrote a query create an unnormalized view of the data.

CMPT 355: Theory and Application of Databases 20

ETL for Assignment 4

SELECT
…
FROM Thread
JOIN Category
ON Category.CategoryID = Thread.CategoryID
JOIN Post
ON Post.ThreadID = Thread.ThreadID
JOIN User AS PostAuthor
ON Post.AuthorUserID = PostAuthor.UserID
JOIN Rating
ON Rating.PostID = Post.PostID
LEFT JOIN ThreadTag
ON ThreadTag.ThreadID = Thread.ThreadID
JOIN Tag
ON Tag.TagID = ThreadTag.TagID

CMPT 355: Theory and Application of Databases 21

ETL for Assignment 4
INSERT INTO Thread (ThreadID, ThreadName,
• And then based on that, CreatedOn, CategoryID, AuthorUserID)
SELECT DISTINCT topic_id, topic, topic_date,
you wrote INSERT INTO … category_id, user_id
SELECT queries to insert FROM temp
the data. GROUP BY topic_id HAVING MIN(post_id);

• Not a lot of transforming INSERT INTO Post (PostID, PostContent,

required (just making use of ThreadID, CreatedOn, AuthorUserID)
default values and NULLs). SELECT DISTINCT post_id, post, topic_id,
post_date, user_id
FROM temp;
• This is almost the reverse INSERT INTO Rating (PostID, RatedByUserID,
of what you would do for a Rating)
data warehouse. SELECT DISTINCT post_id, user_id_rating, rating
FROM temp
CMPT 355: Theory and Application of Databases 22
ETL Tools

• There are purpose-built ETL tools available for performing similar

migration tasks.

• You can always create

your own scripts, e.g.,
using your Python skills.

CMPT 355: Theory and Application of Databases 23

 https://pokemondb.net/type
ETL Tools

• For example, the Pokemon

TypeChart requires a complicated
transform if we were to do it purely
in SQL.
• Using Python would not be so
difficult.

CMPT 355: Theory and Application of Databases 24

ETL for TypeChart
Initial import and result of converting to CSV

CMPT 355: Theory and Application of Databases 25

ETL for TypeChart
Cleaning up values and column names

CMPT 355: Theory and Application of Databases 26

ETL for TypeChart
Importing into SQLite DB

CMPT 355: Theory and Application of Databases 27

ETL for TypeChart

• I want to load the values into this table

CREATE TABLE TypeChart (

AttackingType TEXT NOT NULL,
DefendingType TEXT NOT NULL,
Multiplier NUMERIC,
PRIMARY KEY (
AttackingType,
DefendingType
)
)

CMPT 355: Theory and Application of Databases 28

ETL for TypeChart
import sqlite3

conn = sqlite3.connect("typechart.db")
cur = conn.cursor()

def fetch_types():
cur.execute("SELECT Attacking FROM tmp")
results = cur.fetchall()

return [r[0] for r in results]

def transform():
types = fetch_types()

for typeAtk in types:

for typeDef in types:
cur.execute("SELECT " + typeAtk + " FROM tmp WHERE Attacking = '" + typeDef + "'")
multiplier = cur.fetchone()[0]

cur.execute("""INSERT INTO TypeChart (AttackingType, DefendingType, Multiplier)

VALUES (?, ?, ?)""", (typeAtk, typeDef, multiplier))
conn.commit()

transform()

CMPT 355: Theory and Application of Databases 29

ETL for TypeChart
The loaded data

• I get 324 rows as expected. (18 types x 18 types)

CMPT 355: Theory and Application of Databases 30

 Textbook Section 32.4
Dimension Modelling

• You can structure the data in a “flatter” way that removes need for
complex JOINs or is more accessible to end-users of the OLAP system
• Every dimensional model is composed of
• One table with many foreign keys making up a composite PK, called the fact
table.
• Many tables, each with a simple surrogate PK, called dimension tables.
• This forms a “star-like” structure, and so is called a star schema.

CMPT 355: Theory and Application of Databases 31

 “DreamHome” database logical design from
the textbook (Figure 17.9)

CMPT 355: Theory and Application of Databases 32

 Star schema (dimensional
model) for property sales
(Textbook Figure 32.2)

• Unnormalized
• One join to get any
dimension we might be
interested in

CMPT 355: Theory and Application of Databases 33

Chinook Database
Logical DB Diagram

CMPT 355: Theory and Application of Databases 34

Chinook Database
Star Schema for Track Sales

CMPT 355: Theory and Application of Databases 35

 Text Section 33.1
OLAP Analyses

• Regardless of whether a data warehouse is used, “OLAP” refers to

specific types of analytical queries.
• “The dynamic synthesis, analysis, and consolidation of large volumes of
multidimensional data.” - Textbook, Section 33.1
• These queries involve multi-dimensional views of aggregate data.
• Not just basic navigation and browsing (“slicing and dicing”), but more
complex analyses involving time or multiple dimensions.
• E.g., “how many tracks were sold of each genre, by region, since 2009.”

CMPT 355: Theory and Application of Databases 36

OLAP Cubes

• Traditionally, in a 2D table you would have two axes, and then values.
• OLAP cubes extend this to many more dimensions

• This is best demonstrated through a pivot table.

• For demonstration purposes, I’m just going to work with a table of unnormalized
data.

CMPT 355: Theory and Application of Databases 37

Pivot Table Data
Chinook Database

CREATE VIEW "Sales" AS

SELECT
II."UnitPrice" AS "UnitPrice",
II."Quantity" AS "Quantity",
II."UnitPrice" * II."Quantity" AS "ExtendedPrice",
G."Name" AS "GenreName",
MT."Name" AS "MediaType",
AL."Title" AS "AlbumTitle",
AR."Name" AS "ArtistName",
C."City", C."State", C."Country",
"InvoiceDate"
FROM "InvoiceLine" AS II
JOIN "Invoice" AS I ON I."InvoiceId" = II."InvoiceId"
JOIN "Track" AS T ON T."TrackId" = II."TrackId"
JOIN "Album" AS AL ON AL."AlbumId" = T."AlbumId"
JOIN "Artist" AS AR ON AR."ArtistId" = AL."ArtistId"
JOIN "Genre" AS G ON G."GenreId" = T."GenreId"
JOIN "MediaType" AS MT ON MT."MediaTypeId" = T."MediaTypeId"
JOIN "Customer" AS C ON C."CustomerId" = I."CustomerId"

CMPT 355: Theory and Application of Databases 38

Pivot Table Data

CMPT 355: Theory and Application of Databases 39

Pivot Table

CMPT 355: Theory and Application of Databases 40

Accomplishing this with SQL

• Goal: Write a query that performs the same calculations and

groupings as in the example.
• Using:
“Year” for the columns dimension,
“Country” for the rows dimension,
“Quantity” for the measure
• We can do this with multiple GROUP BYs and UNIONs

CMPT 355: Theory and Application of Databases 41

Accomplishing this with SQL

SELECT
"Country",
date_part('year', "InvoiceDate") AS "Year",
SUM("Quantity")
FROM "Sales"
GROUP BY "Country", "Year"

CMPT 355: Theory and Application of Databases 42

Accomplishing this with SQL

SELECT
"Country",
date_part('year', "InvoiceDate") AS "Year",
SUM("Quantity")
FROM "Sales"
GROUP BY "Country", "Year"

We still need:
• Subtotals for country, year
• Grand total

CMPT 355: Theory and Application of Databases 43

Accomplishing this with SQL
Adding subtotals and grand total

Country, Year SELECT SELECT Sub-Total for

combination "Country", "Country", each Country
date_part('year', "InvoiceDate") NULL,
AS "Year", SUM("Quantity")
SUM("Quantity") FROM "Sales"
FROM "Sales" GROUP BY "Country"
GROUP BY "Country", "Year"
UNION
UNION
SELECT Sub-Total for
SELECT NULL, each Year
Grand Total NULL, date_part('year', "InvoiceDate")
NULL, AS "Year",
SUM("Quantity") SUM("Quantity")
FROM "Sales" FROM "Sales"
GROUP BY "Year"
UNION
CMPT 355: Theory and Application of Databases 44
GROUPING SETS (SQL-99)

• That was tedious.

• Wouldn’t it be nice if we could do that in one query?
• Just use grouping sets!
• In general… SELECT "column1", "column2"
FROM "table"
SELECT "column1", "column2" GROUP BY "column1", "column2"
FROM "table"
GROUP BY Is shorthand for… UNION
GROUPING SETS (
("column1",) SELECT "column1", NULL
("column1", "column2") FROM "table"
) GROUP BY "column1"

CMPT 355: Theory and Application of Databases 45

GROUPING SETS (SQL-99)

• So we can write a much simpler query…

SELECT
"Country",
date_part('year', "InvoiceDate") AS "Year",
SUM("Quantity")
FROM "Sales"
GROUP BY GROUPING SETS (
(),
("Country"),
("Year"),
("Country", "Year")
)

CMPT 355: Theory and Application of Databases 46

CUBE

• This type of OLAP Cube analysis is so common that there’s a CUBE

shorthand for GROUPING SETS
SELECT SELECT
"Country", "Country",
date_part('year', "InvoiceDate") date_part('year', "InvoiceDate")
AS "Year", AS "Year",
SUM("Quantity") SUM("Quantity")
FROM "Sales" FROM "Sales"
GROUP BY GROUPING SETS ( GROUP BY CUBE ("Country", "Year")
(),
("Country"),
("Year"),
("Country", "Year")
)

CMPT 355: Theory and Application of Databases 47

 https://www.postgresql.org/docs/10/
CUBE queries-table-expressions.html#QUERIES-GROUPING-SETS

In General

• In general…
GROUPING SETS (
( a, b, c ),
( a, b ),
Is equivalent to ( a, c ),
CUBE ( a, b, c )
( a ),
( b, c ),
( b ),
( c ),
( )
)

CMPT 355: Theory and Application of Databases 48

 https://www.postgresql.org/docs/10/
ROLLUP queries-table-expressions.html#QUERIES-GROUPING-SETS

• Another shorthand notation for GROUPING SETS is ROLLUP

GROUPING SETS (
( e1, e2, e3, ... ),
ROLLUP ( e1, e2, e3, ... ) ...
( e1, e2 ),
Is equivalent to ( e1 ),
( )
)

CMPT 355: Theory and Application of Databases 49

Window Functions (SQL:2003)

• Another OLAP-related feature in SQL are window functions.

• Window functions allow you to show the results of various
calculations at the row-level.

CMPT 355: Theory and Application of Databases 50

 http://www.postgresqltutorial.com/
Window Functions (SQL:2003) postgresql-window-function/

• Window functions can be categorized into:

• Value window functions • You can also use the aggregate
• FIRST_VALUE, LAST_VALUE functions as window functions
• LAG, LEAD • AVG, COUNT, MAX, MIN, SUM
• Ranking window functions
• ROW_NUMBER, RANK, DENSE_RANK
• They use the syntax…
window_function(arg1, arg2,..) OVER (
PARTITION BY expression
ORDER BY expression Partition is optional
)

CMPT 355: Theory and Application of Databases 51

 http://www.postgresqltutorial.com/
Window Functions (SQL:2003) postgresql-window-function/
http://www.sqlitetutorial.net/sqlite-
window-functions/

Name Description
Compute the rank for a row in an ordered set of rows with no gaps in rank
DENSE_RANK
values.
RANK Assign a rank to each row within the partition of the result set.
FIRST_VALUE Get the value of the first row in a specified window frame.
Provide access to a row at a given physical offset that comes before the
LAG
current row.
LAST_VALUE Get the value of the last row in a specified window frame.
LEAD Provide access to a row at a given physical offset that follows the current row.
Assign a sequential integer starting from one to each row within the current
ROW_NUMBER
partition.

CMPT 355: Theory and Application of Databases 52

Example
Top Ranked Country by Total Sales

• It’s trivial to create a query to list the countries in order by sales

SELECT
"Country",
SUM("ExtendedPrice") AS "Sales"
FROM "Sales"
GROUP BY "Country"
ORDER BY "Sales" DESC

• But what if we wanted to rank them?

• USA #1
• Canada #2
• etc.

CMPT 355: Theory and Application of Databases 53

Example
Top Ranked Country by Total Sales

• Use the RANK() window function.

SELECT
"Country",
SUM("ExtendedPrice") AS "Sales",
RANK() OVER (ORDER BY SUM("ExtendedPrice") DESC)
FROM "Sales"
GROUP BY "Country"
ORDER BY "Sales" DESC

CMPT 355: Theory and Application of Databases 54

Example
Country rank by year

• Let’s say our query was a bit more complicated.

• Each country’s sales by year
SELECT
"Country",
SUM("ExtendedPrice") AS "Sales",
DATE_PART('year', "InvoiceDate") AS "Year"
FROM "Sales"
GROUP BY "Country", "Year"
ORDER BY "Sales" DESC

• We now want to know the country’s rank over each year

CMPT 355: Theory and Application of Databases 55

Example
Country rank by year

• Now we need PARTITION BY

SELECT
"Country",
SUM("ExtendedPrice") AS "Sales",
DATE_PART('year', "InvoiceDate")
AS "Year",
RANK() OVER (
PARTITION BY
DATE_PART('year', "InvoiceDate")
ORDER BY SUM("ExtendedPrice") DESC
)
FROM "Sales"
GROUP BY "Country", "Year"
ORDER BY "Sales" DESC

CMPT 355: Theory and Application of Databases 56

Example
Rank by country, year, subtotals

• We can combine the CUBE with RANK to get even more…

SELECT
"Country",
SUM("ExtendedPrice") AS "Sales",
DATE_PART('year', "InvoiceDate")
AS "Year",
RANK() OVER (
PARTITION BY
DATE_PART('year', "InvoiceDate")
ORDER BY SUM("ExtendedPrice") DESC
)
FROM "Sales"
GROUP BY CUBE("Country", "Year")
ORDER BY "Sales" DESC

CMPT 355: Theory and Application of Databases 57

Example: Distance from X,Y,Z
From Assignment 2’s database

• In Assignment 2, there was a “movement” table

CMPT 355: Theory and Application of Databases 58

Example: Distance from X,Y,Z
From Assignment 2’s database

• In Assignment 2, there was a “movement” table

• I am interested in the total distance travelled for each logNavigationID
• From one row, we need to access the row after (or before)
• Calculate the distance from one row to another
• Then calculate the total over the logNavigationID

CMPT 355: Theory and Application of Databases 59

Example: Distance from X,Y,Z
From Assignment 2’s database

• Let’s start with accessing the adjacent row…

SELECT
movement.lognavigationid,
movement.posx,
movement.posy,
movement.posz
FROM movement

CMPT 355: Theory and Application of Databases 60

Example: Distance from X,Y,Z
From Assignment 2’s database

• Let’s start with accessing the adjacent row…

I’ll use the LAG function to get the data from the previous row…

LEAD (expression [, offset]) OVER (PARTITION BY ___ ORDER BY ___)

CMPT 355: Theory and Application of Databases 61

Example: Distance from X,Y,Z
From Assignment 2’s database

SELECT
movement.lognavigationid,
movement.posx,
LAG(movement.posx, 1) OVER (
PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posx,
movement.posy,
LAG(movement.posy, 1) OVER (
PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posy,
movement.posz,
LAG(movement.posz, 1) OVER (
PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posz
FROM movement

CMPT 355: Theory and Application of Databases 62

 Example: Distance from X,Y,Z
From Assignment 2’s database

SELECT
lognavigationid,
SQRT((m.posx-m.prev_posx)^2 + (m.posy-m.prev_posy)^2 + (m.posz-m.prev_posz)^2) AS distance
FROM
(
SELECT
movement.logmovementid,
movement.lognavigationid,
movement.posx,
LAG(movement.posx, 1)
OVER (PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posx,
movement.posy,
LAG(movement.posy, 1)
OVER (PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posy,
movement.posz,
LAG(movement.posz, 1)
OVER (PARTITION BY lognavigationid ORDER BY logmovementid ASC) AS prev_posz
FROM movement
) AS m

CMPT 355: Theory and Application of Databases 63

Example: Distance from X,Y,Z

Now calculate SUM for each logNavigationID

SELECT
lognavigationid,
SUM(SQRT(
(m.posx-m.prev_posx)^2
+ (m.posy-m.prev_posy)^2
+ (m.posz-m.prev_posz)^2)) AS distance
FROM
(
…
) AS m
GROUP BY lognavigationid

CMPT 355: Theory and Application of Databases 64

Example: Distance from X,Y,Z

What I was really interested in was between-group conditions

SELECT
P."condition",
SUM(SQRT((
m.posx-m.prev_posx)^2
+ (m.posy-m.prev_posy)^2
+ (m.posz-m.prev_posz)^2)) AS distance
Group 0 travelled a greater distance
FROM
(more error in completing the navigation task)
(
...
) AS M
JOIN log_navigation LN ON LN.lognavigationid = M.lognavigationid
JOIN participant P ON P.participantid = LN.participantid
GROUP BY P."condition"

CMPT 355: Theory and Application of Databases 65

Summary

• OLTP is what we’ve been doing in this course.

• OLAP is all about analytical, read-only queries
• OLAP can be supported by the creation of a dedicated data
warehouse.
• Data warehouses are populated using ETL: Extract, Transform, Load
• The structure of a data warehouse can be modelled by a “Star Schema,” a
single fact table with many dimension tables
• In SQL, OLAP is facilitated by two features
• GROUPING SETS
• Window functions

CMPT 355: Theory and Application of Databases 66