Apex Project

Retail & E-Commerce Inventory AI Pipeline

BITS Pilani · BS Program · Apex Project · Trimester 3

---

Student Details

Field	Details
Student	Samarth Rajesh Lad
Problem ID	P4 - Retail and E-Commerce: Conversational Inventory AI Pipeline
Program	BITS Pilani BS Program, Trimester 3
Instructor	Deepak Bahuguna

---

About

Raw retail transaction data is rarely clean. Missing values, wrong data types, and inconsistent formatting quietly undermine any analysis built on top of them.

This project builds a 14-step end-to-end data cleaning pipeline for a retail transactions dataset sourced from Kaggle, transforming messy raw data into a clean, structured, analysis-ready dataset.

	Raw Dataset	Clean Dataset
Rows	12,575	11,362
Columns	11	17
Missing Values	7,229	0
Duplicate Rows	0	0

---

Dataset

Retail Store Sales - Dirty for Data Cleaning

Source: Kaggle

Known issues in raw data:

• 1,213 missing Item names
• 609 missing Price Per Unit values
• 604 missing Quantity values
• 604 missing Total Spent values
• 4,199 missing Discount Applied values
• Transaction Date stored as plain text
• Quantity stored as float instead of integer
• Discount Applied stored as text instead of boolean

---

Pipeline Steps

Step	Activity
1	Load dataset and inspect structure
2	Record before snapshot
3	Remove rows with missing Item names
4	Check missing Quantity and Total Spent
5	Fill missing Price Per Unit with category median
6	Fill missing Discount Applied with False
7	Convert Transaction Date to datetime format
8	Convert Quantity from float to integer
9	Standardize Category and Item text formatting
10	Label encode categorical columns
11	Apply MinMaxScaler to numeric columns
12	Generate before/after data quality report
13	Produce three analysis charts
14	Export clean_retail_store_sales.csv

---

Key Findings

From the cleaned dataset of 11,362 rows:

• Butchers was the top category by total sales (~200,000)
• Online and In-store channels contributed almost equally
• Cash was the most used payment method (~510,000 in sales)

---

Repository Files

File	Description
Apex_Project_Samarth_Lad.ipynb	Main pipeline notebook
Apex_Project_Report_Samarth_Lad.pdf	Full project report
GenAI_Prompt_Log_Samarth_Lad.pdf	GenAI usage log

---

Tools and Libraries

Tool	Version	Purpose
Python	3.10	Core programming language
Pandas	2.2.2	Data loading, cleaning, transformation
NumPy	2.0.2	Numeric operations
scikit-learn	1.6.1	LabelEncoder, MinMaxScaler
Matplotlib	3.10.0	Bar charts and visualizations
Seaborn	0.13.2	Styled visualizations
Google Colab	Latest	Development environment

---

_{BITS Pilani BS Program · Apex Project · Trimester 3 · P4}