SQL for Data Science

Predict
 Contents

Problem Context/Domain

Predict Rules + Instructions

Student Starter Pack

Important Packages

Data ERD Guide

Predict FAQs

2
Problem Context/Domain - Retail (Online Retailing Business)

Problem Statement:
The Bhejane Trading store is an online retailer specializing in the sale of covid-related essential items. As a
consultant hired by the company, you have been tasked with the objective of normalizing the database of the
store’s inventory management system.
You are provided with an unnormalised database, and are expected to normalise it's contents to bring it into 3rd
Normal Form (3NF). The database has 2 tables (products and transactions) which are summarised here.

Deliverables:
After having normalised the DB, you will be required to answer several multiple-choice questions which test
your completed work, and your practical SQL skills gained in the course.

NB: The following deliverables must also be uploaded to Athena.

- The notebook should also contain queries used to answer the MCQ.
- A SQLite ‘.db’ database file of the normalised database.

© Explore Data Science Academy

Predict Rules + Instructions

● This project is an individual project; your work needs to reflect your understanding of the course content.

● You are free to share ideas with colleagues and classmates, however, you are not allowed to share your code, solutions, or
submissions with any other individual. Plagiarism will not be tolerated.

● You are required to submit all of the code you use to both normalise the given dataset, and to answer the related MCQ
assessment. Submit your completed starter notebook, along with any other material as a zipped file under the ‘Upload Predict
File’ tab on Athena.

● The official due date of the Predict will be displayed on Athena. No submissions after 23:59 on this date will be accepted for
marking.

© Explore Data Science Academy

Student Starter Pack - Getting You on the Right Track

In order to help you get your bearings within the Predict, we’ve prepared a ‘starter pack’ which contains
essential material to guide your work. This material includes:

● Base notebook: A Jupyter notebook containing code and instructions to begin work on the Predict.
Continue developing this file to use for final solution submission to Athena.
● The unnormalised data: Two .csv files containing the unnormalised data.
○ ‘bhejane_covid_essentials_Products.csv’
○ ‘bhejane_covid_essentials_Transactions.csv’
● A description of the various data fields found in the database.

© Explore Data Science Academy

Making Sense of our Queries

Within this Predict we’ll be writing a lot of SQL statements. In order to make your SQL queries more
human-readable and to help you along, we will install an ipython-sql package to assist with syntax highlighting.

● Install the ipython-sql package by entering the following command into your terminal:
pip install ipython-sql.
● Now you can use the %%sql magic command at the start of each cell when writing your SQL queries and
the syntax will be highlighted.

© Explore Data Science Academy

Detailing the Data - Original Database Tables

To help familiarise yourself with the data

in the original database, we provide the
following ERD - showing the various
fields for the Products and
Transactions tables respectively.

You are required to use the principles of

database normalisation to transform
these tables into the 3NF schema.
Subsequent slides will detail the
normalization process

*NB: Be wary of handling NULL values

in the dataset

© Explore Data Science Academy

Detailing the Data - 1NF Entity Relationship Diagram

Throughout the Predict you will be

given the target ERD for each
normalization step.

To the right is the ERD sketch for the

1st Normal Form to get you started.

Pay attention to field attributes such as

data types, primary keys, composite
keys, foreign keys and relationships
that exist amongst them

© Explore Data Science Academy

Detailing the Data - 2NF Entity Relationship Diagram

You are encouraged to use the AUTOINCREMENT property when creating new
fields that are going to be used as primary keys.

© Explore Data Science Academy

Detailing the Data - 3NF Entity Relationship Diagram

Hint:
As you progress through
the different normal forms
you may find it easier
populate the current
normal forms using the
previous normal forms

© Explore Data Science Academy

Predict-related FAQs

This page will be updated periodically with common predict-related questions which may arise during the
Sprint. Consider consulting this space before asking your course facilitator a question.

Considerations to keep in mind when completing the predict, before answering the predict questions.

1. The aim of the predict is to understand and implement normalization on the dataset provided. This includes,
understanding separation of entities (tables which serve a single purpose), maintaining relationships and
enforcing normalization through data integrity.

2. Following the normalization process is an important step to follow in order to be able to answer the predict
questions effectively.

3. Having an understanding of your problem and data can be very helpful in guiding your thinking to solve a
problem. At each stage of the normalization it is suggested that you take some time to reflect on what changes
were made from the previous normal form and understand why transformations were made.

© Explore Data Science Academy

Predict-related FAQs

I am getting the following error - ModuleNotFoundError: No module named 'ipython-sql'; What should I do?
● Please make sure that you have installed the ipython-sql using the following command: pip install ipython-sql

I cannot make changes to my table creation code, I get the following error everytime I try - OperationalError:
table <TableName> already exists
● You are advised to first drop the old table before re-creating the table with your new changes
○ DROP TABLE IF EXISTS [TableName];

● You can drop and create the tables as many time as you want, just remember to keep the table naming
convention consistent with the ERD sketches that are provided.

What does ‘PK’ and ‘FK’ stand for when looking at the ERD sketches?
● PK: Primary Key
● FK: Foreign Key

© Explore Data Science Academy

Predict-related FAQs

I am constantly getting errors and debugging is a nightmare

● SQL by nature requires one to be pedantic - so pay special attention to syntax and formatting. If your SQL queries
generally look like the below - may the debugging gods be with you…

● SQL doesn’t have any formatting rules (such as indentation in python), so it will allow you to run the above query
with no issues at all. It is however recommended you practise good SQL hygiene and stay away from this practice.
Although there is no book of all truths for SQL formatting, it should generally take the following form:

© Explore Data Science Academy

Predict-related FAQs

How can I compare my normalised database to the reference ERD diagrams?

● ERAlchemy is a useful package for viewing relationship diagrams within Jupyter
ERAlchemy requires GraphViz to generate the graphs and Python. Both are available for Windows, Mac and Linux.

● Within a Jupyter codecell, execute the render_er Python function to see your relationship diagram

● Or be more specific on the tables you want to include in the output

© Explore Data Science Academy