FMCG A/B Test Sales Analysis

Author: Agnes Offenbek
Role: Business Analyst | Junior Data Scientist
Project Type: Portfolio Project
Tools: Python, Pandas, Matplotlib, Jupyter Notebook

---

Project Overview

This project demonstrates an A/B test analysis for a fictional FMCG retail scenario.
The goal is to compare sales performance between stores with a discount campaign (treatment group) and control stores.

• Clean, structured data is used from a simulated sales dataset.
• End-to-end analysis is performed in Python (Jupyter Notebook).
• Visualizations are created using Matplotlib.
• Summary statistics and CSV output are generated for further reporting.

---

Folder Structure

fmcg-ab-test-sales-analysis/
│
├── data/
│ ├── raw/ # Original dataset
│ └── processed/ # Summary tables and processed outputs
│
├── notebooks/
│ └── 01_ab_test_analysis.ipynb # Main analysis notebook
│
├── src/
│ └── data_generation.py # Script to generate example dataset
│
├── README.md
└── requirements.txt # Python dependencies

---

How to Run

1. Clone the repository

git clone https://github.com/aoffenbek/fmcg-ab-test-sales-analysis.git
cd fmcg-ab-test-sales-analysis

2. Create a virtual environment (optional but recommended)

python -m venv venv source venv/Scripts/activate # Windows

3. Install dependencies

pip install -r requirements.txt

4. Run the notebook

Open Jupyter Notebook: Navigate to notebooks/01_ab_test_analysis.ipynb and run all cells.

---

Project Steps

1. Data Generation / Loading – generate or load raw sales data.
2. Grouping Stores – assign stores to discount or control group.
3. Exploratory Data Analysis – calculate daily sales, averages, and visualize.
4. Summary Statistics – compute mean, standard deviation, and export results.
5. Visualization – bar charts comparing discount vs control groups.
6. Output – save summary tables for further business analysis.

---

Example code snippets

Load data

import pandas as pd

Load raw sales data

sales_df = pd.read_csv('data/raw/sales_data.csv')

Assign Groups for A/B Test

Assign stores to discount or control group

sales_df['group'] = sales_df['store'].apply(lambda x: 'discount' if int(x.split('_')[1]) <= 10 else 'control')

Calculate Summary Statistics

Average daily sales per group

summary = sales_df.groupby('group')['daily_sales'].agg(['mean','std']).reset_index() summary.to_csv('data/processed/ab_test_summary.csv', index=False)

Simple Visualization import matplotlib.pyplot as plt

plt.bar(summary['group'], summary['mean'], yerr=summary['std'], color=['skyblue','orange']) plt.title('Average Daily Sales: Discount vs Control') plt.ylabel('Average Sales') plt.show()

---

Skills Demonstrated

• Data Analysis & Cleaning: Pandas, SQL basics
• Visualization: Matplotlib charts for A/B comparison
• Statistical Thinking: Basic summary statistics for A/B testing
• Portfolio-Ready Workflow: Folder structure, modular code, reproducibility
• Documentation & Reporting: Clear README and notebook explanation

---

Requirements

All Python dependencies are listed in requirements.txt. Install them with:

pip install -r requirements.txt

Minimum packages:

pandas>=1.5.0 matplotlib>=3.7.0 numpy>=1.25.0 jupyter>=1.0.0

Optional (for SQL queries):

sqlalchemy>=2.0

---

Next Steps / Extensions

• Apply hypothesis testing (t-tests) to confirm statistical significance.
• Integrate time-series analysis for trends over weeks/months.
• Simulate realistic FMCG datasets with multiple product categories.
• Explore Python visualization libraries (Seaborn, Plotly) for interactive dashboards.