Giters

erdosn / cv2-mod2-section09-sql-lesson

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Table of Contents

Lesson Structure

Questions

Outcomes

YWBAT

  • perform a query that is ordered by a value
  • load query results as a dataframe
  • build functions to perform queries
  • execute a query using multiple joins

Objectives

  • write an order by query on a given table
  • convert the customers table to a dataframe using pandas and sqlite3
  • compare and contrast using a dataframe vs using the results as a list, from a query
  • write a single join query
  • write a multiple join query

Outline

  • Take Questions
  • Warm Up
  • Connect to our sqlite db
  • Compare and contrast reading a query as a list vs reading a query as a dataframe
  • Practice various queries
  • Complete a Join Query
  • Complete a query using multiple Joins
  • wrap up

(5 min) Warm Up

Given the following 'Employees' table, write a query that completes the following task. Send your query to me in a private chat here in zoom.

Expected Result:

  • IdNum, LName, FName and Salary
  • Order it by salary, starting with the highest salary

Solution 
SELECT IdNum, LName, FName, Salary
FROM Employees
ORDER BY Salary DESC

Demonstrating SQLite & Python

import pandas as pd
import numpy as np

import sqlite3

import matplotlib.pyplot as plt

Connecting to our sqlite db using sqlite3

conn = sqlite3.connect('data.sqlite')
cursor = conn.cursor()

Listing the tables in our db

# table_names to be a list of my table_names
query = "SELECT name FROM sqlite_master WHERE type='table';"
res = cursor.execute(query).fetchall()
print(res) # Notice these are tuples, let's extract only the names in the next cell
table_names = [r[0] for r in res]
table_names

Now select everything from the employees table just to get a feel for it

query = 'select * from employees;'
res = cursor.execute(query).fetchall()
res[:2]

Now let's load this into a dataframe using the .read_sql method, we'll use some of the same components from above.

query = 'select * from employees;'
df = pd.read_sql(query, conn)
df.head() # Much better and more readable!

(3 min) What are the pros and cons of loading a sql query into a dataframe?

  • Pros

  • Cons

def load_df(table_name=None, conn=None):
    query = 'select * from {}'.format(table_name)
    df = pd.read_sql(query, conn)
    return df

A throwback favorite

d = {} # table_name: dataframe of table
for table_name in table_names:
    d[table_name] = load_df(table_name, conn)

WHY SHOULD YOU ALWAYS SPECIFY YOUR PRIMARY KEYS!?

[Discussion here]

Exercises

Query Exercise 1

Create the following table using the customers table and the orders table

# First look at the customers dataframe 


# Then look at the orders dataframe
Query 1 - Solution 
    query = """
    SELECT c.customerName, c.phone, c.state, o.orderDate FROM
    customers as c
    JOIN orders as o using (customerNumber);
    """
query = """SELECT
        """
join_1 = pd.read_sql(query, conn)
join_1.head()

Query Exercise 2

Create a query that results in a table with the every customer number for an employee along with the employee name and number.

Ex:

# Work here
query = """
        SELECT
        """

pd.read_sql(query, conn).head()
Query 2 - Solution 
query = """
    select e.firstname, e.lastname, e.employeenumber, c.customerNumber
    from employees as e
    join customers as c
    on e.employeenumber = c.salesrepemployeenumber;"""
# Now let's take this and incorporate the next table orders
orders_df.head(1)

Query Exercise 3

Create a query that results in a table with every order number for every customer number for an employee along with the employee name and number.

Ex:

# Work here
query = """
        SELECT
        """

pd.read_sql(query, conn).head()
Query 3 - Solution 
query = """
    select e.firstname, e.lastname, e.employeenumber, c.customerNumber, o.orderNumber
    from employees as e
    join customers as c
    on e.employeenumber = c.salesrepemployeenumber
    join orders as o
    on o.customerNumber = c.customerNumber;"""

Query Exercise 4

Write a query that results in the following:

Employee First and Last Name, order number and each product code for that order.

There should be a row for each product code

Screenshot:

# work here
query = """
        select
        """

pd.read_sql(query, conn).head(2)
Query 4 Solution 
query = """
    select e.firstname, e.lastname, o.orderNumber, od.productCode
    from employees as e
    join customers as c
    on e.employeenumber = c.salesrepemployeenumber
    join orders as o
    on o.customerNumber = c.customerNumber
    join orderdetails as od
    on od.orderNumber = o.orderNumber;"""
    ```
        
</details>

## Query Exercise 5

Write a query that results in the following:

Employee First and Last Name and each product name they sold, add an alias to match the image below

*There should be a row for each product name*

Screenshot:

<img src="images/query5.png" width="500"/>


```python
# work here
query = """
        select
        """

pd.read_sql(query, conn).head(2)
Query 5 Solution 
query = """
    select e.firstname, e.lastname, p.productName
    from employees as e
    join customers as c
    on e.employeenumber = c.salesrepemployeenumber
    join orders as o
    on o.customerNumber = c.customerNumber
    join orderdetails as od
    on od.orderNumber = o.orderNumber
    join products as p
    on p.productCode = od.productCode;"""
    ```
        
</details>

## Great! It's working. Let's add some group by to remove duplicates!

**Add aliases to result in the following**

<img src="images/query5.png" width="500"/>


```python
# remember the alias' from above

query = '''SELECT ...'''

final_df = pd.read_sql(query, conn)
final_df.head(2)
Final Query Solution 
query = """
        select e.firstname as fn, e.lastname as ln, p.productName as pn
        from employees as e
        join customers as c
        on e.employeenumber = c.salesrepemployeenumber
        join orders as o
        on o.customerNumber = c.customerNumber
        join orderdetails as od
        on od.orderNumber = o.orderNumber
        join products as p
        on p.productCode = od.productCode
        group by fn, ln, pn;"""

About

Languages

Language:Jupyter Notebook 100.0%