SQL Database

Data Science Bootcamp

Why should we learn databases?

Databases

Spreadsheets Dashboar
d
What can databases do?
Some databases also
have data visualization
features

Store Analyze Present

SQL is a fundamental skill for data analyst

SQL is easy to learn, Quick win!

Average salary in USA
Data Analyst with SQL Skills Salary | PayScale

Convert to T HB
* 30 / 2.5

You can find your net

worth using this free tool
There are many versions of SQL

We’ll teach PostgreSQL too :)

SQLite is a standard
language (pure)

State of Data Science and Machine Learning 2020 | Kaggle

We have been using SQL for 50 years

Structured Query Language

verb. Ask a question about our data

How database work?

Query

Result

Every company has databases

(99.99%)
 What database looks like?
You are already familiar with
database technology

Customers Sales Invoices Products

At the heart of SQL is selecting the data you want

Select columns

Filter rows
ER Diagram Entity

Relationship

SQLite Tutorial - An Easy Way to Master SQLite Fast

Primary and foreign keys

Table: Albums

Column Type

Albumid Integer

Table: Artists Title Text

Column Type Artistid Integer

Artistid Integer

Name Text
SQL clauses we use in our data analyst role

Clauses What it does?

SELECT Select columns

FROM From table

JOIN Join multiple tables

WHERE Filter data

Aggregate Functions AVG SUM MIN MAX COUNT

GROUP BY Group by statistics

HAVING Filter groups

ORDER BY Sort data

The first part will cover all the SQL basics :)
Select all columns

SELECT * FROM customers;

Select all columns

Table name

SELECT * FROM customers;

Upper case Close with ;

Select specific columns

SELECT
firstname,
lastname,
email,
country
FROM customers;
Choose specific columns
Filter rows with where clause

condition
SELECT *
FROM customers
WHERE country = ‘USA’;

SELECT *
FROM customers
WHERE country IN (‘USA’, ‘Canada’, ‘United Kingdom’);
Review join types

Inner Join Left Join Inner and Left Join contribute

around 90-95% of our work

Right Join Full Join

Inner join

Table 1 Table 2 Result Set

PK_ID Name FK_ID Major PK_ID Name Major

1 David 1 Econ 1 David Econ

2 John 2 Econ 2 John Econ

+ =
3 Marry 5 Data 5 Kevin Data

4 Anna 12 Engineer

5 Kevin 35 Mkt
Left join

Table 1 Table 2 Result Set

PK_ID Name FK_ID Major PK_ID Name Major

1 David 1 Econ 1 David Econ

2 John 2 Econ 2 John Econ

+ =
3 Marry 5 Data 3 Marry NULL

4 Anna 12 Engineer 4 Anna NULL

5 Kevin 35 Mkt 5 Kevin Data

Full join

Table 1 Table 2 Result Set

PK_ID Name FK_ID Major PK_ID Name Major

1 David 1 Econ 1 David Econ

2 John 2 Econ 2 John Econ

+ =
3 Marry 5 Data 3 Mary NULL

4 Anna 12 Engineer 4 Anna NULL

5 Kevin 35 Mkt 5 Kevin Data

12 NULL Engineer

35 NULL Mkt
Join example

SELECT A.*, B.* This query joins two tables -

FROM customers A customers and invoices
JOIN invoices B
ON A.customerid = B.customerid;
Join example

SELECT
A.*,
B.*, It’s easy to join more than two
C.*, tables (4-5 tables is quite normal)
D.*
FROM table1 A
JOIN table2 B ON A.id = B.id
JOIN table3 C ON B.id = C.id
JOIN table4 D ON C.id = D.id;
Warning! The following slides are advanced
SQL topics. Be prepared and hope you enjoy.
Repl.it
Repl.it online editor
Essential command lines

Command What it does?

.help เรียกดูชื่อ sql commands ทั้งหมด

.open เปิดไฟล์ database

.read อ่านไฟล์ sql script

.mode เปลี่ยน mode การแสดงผล

.header แสดงชื่อ column ใน terminal

.table แสดงชื่อ tables ใน database

.schema แสดง schema ของตารางทั้งหมด

.import นําเข้า csv file เป็น table

We can also run SQL in terminal

We call this “T ERMINAL”

Join syntax

Alias or shorter name

SELECT A.*, B.*
FROM customers A
JOIN invoices B
ON A.customerid = B.customerid;

PK = FK
Join using

SELECT A.*, B.*

FROM customers A
JOIN invoices B
USING (customerid);

Using if the column names in

both tables are the same
Join more than one column

SELECT A.*, B.*

FROM tableA A
JOIN tableB B
ON A.customerid = B.customerid
AND A.country = B.country;

use AND to add more columns to join

Review CASE syntax

Condition Value if True

SELECT
CASE
WHEN company IS NULL THEN ‘End Customers’
ELSE ‘Corporate’
END AS segment
FROM customers;

Value if False
Case + Aggregate Functions

SELECT
CASE
WHEN company IS NULL THEN ‘End Customers’
ELSE ‘Corporate’
END AS segment,
COUNT(*) AS N
FROM customers
GROUP BY 1;
Count customers in each segment
Intro to subqueries

Outer Query

Inner Query
Inner run first

SELECT firstname, lastname, country FROM (

SELECT * FROM customers

WHERE country IN (‘USA’, ‘United Kingdom’, ‘Canada’)
)

ORDER BY 3 DESC;
Inner Query
Outer run later

SELECT firstname, lastname, country FROM (

SELECT * FROM customers

WHERE country IN (‘USA’, ‘United Kingdom’, ‘Canada’)
)

ORDER BY 3 DESC;

Outer Query
We often use subqueries in where clause

SELECT * FROM tracks

WHERE bytes = (
SELECT MAX(bytes) FROM tracks
);

Find max bytes

Intro to window functions

Two things you need to know

about window functions

1. they are in SELECT clause

2. they create new columns in
the result table
Aggregate vs. Window functions

Single Value Column

Aggregate Function Window Function

Window function syntax

WINDOW_FUNCTION( ) OVER( PARTITION BY … ORDER BY … )

Function Name Create window

The easiest window function

SELECT
firstname,
lastname,
ROW_NUMBER() OVER() AS rowNum
FROM customers;
Common window functions

Function What it does?

ROW_NUMBER() สร้างคอลัมน์ row number เรียงตั้งแต่ 1 - n

RANK() สร้างคอลัมน์ ranking

DENSE_RANK() สร้างคอลัมน์ ranking

LAG() สร้างคอลัมน์ LAG value (t-1)

LEAD() สร้างคอลัมน์ LEAD value (t+1)

NTILE() สร้างคอลัมน์ segment จับกลุ่ม records

SUM() OVER() สร้างคอลัมน์ผลรวมแบบ running total

Bootcamp Live 02
Intro to Databases with
Website: https://datarockie.com