🏷️ Retail Price Optimization Engine

End-to-end Prescriptive Analytics using Machine Learning

Simulates price elasticity and competitor data to prescribe optimal pricing strategies for revenue maximization.

🎮 Live Demo • Features • Installation • Usage

---

📋 Table of Contents

• Overview
• Features
• Architecture
• Project Structure
• Installation
• Usage
• Testing
• Docker
• Results & Impact

---

🎯 Overview

The Retail Price Optimization Engine is a machine learning solution designed to shift retailers from reactive pricing to Prescriptive Analytics. Unlike traditional forecasting, this system doesn't just predict sales it calculates the Optimal Price Point.

By leveraging Random Forest Regression and other ensemble methods, the system learns the non-linear relationship between Price and Demand (Elasticity), factoring in competitor pricing to recommend strategies that maximize Total Revenue.

Why This Matters

Most retailers still rely on gut feeling or simple cost-plus formulas for pricing. This leads to:

• Leaving money on the table (pricing too low)
• Losing customers (pricing too high)

This engine uses actual data to find that sweet spot where revenue is maximized.

---

✨ Features

🤖 Core Intelligence

• Elasticity Modeling: Captures non-linear demand response to price changes
• Competitor Benchmarking: Real-time tracking of price gaps against competitors
• Revenue Maximization: Simulation algorithm tests 50+ price points per SKU
• Confidence Intervals: Uncertainty estimation for reliable recommendations

📊 Interactive Dashboard

• Single & Batch Optimization: Optimize one product or entire categories
• Real-time Inference: Sub-second prediction for pricing scenarios
• Visual Analytics: Interactive Plotly charts showing Revenue Curves
• Export to Excel/CSV: Download optimization results
• Strategic Insights: Automated recommendations (raise/lower/maintain price)

🧪 Quality Assurance

• 75 Unit Tests: Comprehensive test coverage
• Input Validation: Robust error handling and data validation
• CI/CD Pipeline: Automated testing with GitHub Actions
• Docker Support: Easy deployment with containerization

---

🏗️ Architecture

graph LR
A[Raw Data CSV] --> B(Feature Engineering)
B --> C{ML Model}
C --> D[Price Simulation]
D --> E[Optimization Engine]
E --> F[Streamlit Dashboard]

Loading

Pipeline Steps:

1. Data Ingestion - Load historical sales and competitor pricing data
2. Feature Engineering - Create features like price differences/ratios vs competitors
3. Model Training - Random Forest learns demand response to price changes
4. Price Simulation - Test different price points (50+ scenarios per product)
5. Optimization - Pick the price with highest predicted revenue
6. Dashboard - Visualize everything in the Streamlit app

---

📂 Project Structure

retail-price-optimization/
├── .github/
│   └── workflows/
│       └── test.yml          # CI/CD pipeline
├── data/
│   ├── processed/            # Cleaned data for app demo
│   └── raw/                  # (Gitignored) Raw source data
├── models/
│   └── pricing_model.pkl     # Trained Random Forest model
├── notebooks/
│   ├── 01_eda_pricing_logic.ipynb
│   ├── 02_feature_engineering.ipynb
│   ├── 03_modeling.ipynb
│   └── 04_price_optimization.ipynb
├── tests/
│   ├── test_app.py           # Validation & elasticity tests
│   ├── test_model_training.py # Model tests
│   └── test_utils.py         # Utility function tests
├── .gitignore
├── app.py                    # Main Streamlit Application
├── config.py                 # Centralized configuration
├── model_training.py         # Model training with CV & tuning
├── utils.py                  # Utility functions & validation
├── requirements.txt          # Production dependencies
├── requirements-dev.txt      # Development dependencies
├── Dockerfile                # Container configuration
├── docker-compose.yml        # Docker Compose setup
└── README.md                 # Documentation

Key Files

File	Description
app.py	Main Streamlit dashboard with batch optimization, export
config.py	Centralized configuration with environment variable support
model_training.py	Train models with cross-validation & hyperparameter tuning
utils.py	Validation functions, elasticity calculation, formatting

---

🚀 Installation

Prerequisites

• Python 3.9 or higher
• Git

Quick Start

# Clone the repository
git clone https://github.com/mugnihidayah/retail-price-optimization.git
cd retail-price-optimization

# Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows
# source venv/bin/activate  # Mac/Linux

# Install dependencies
pip install -r requirements.txt

# Run the app
streamlit run app.py

The app will open in your browser at http://localhost:8501.

---

💻 Usage

Running the Application

streamlit run app.py

How to Use

1. Open the app in your browser
2. Select a product category from the sidebar
3. Choose Single Product or Batch Optimization mode
4. Click "Run AI Optimization"
5. Review results and download recommendations

What You'll See

• Revenue Curve: How revenue changes at different price points
• Optimal Price: AI-recommended price with confidence interval
• Revenue Uplift: Potential revenue increase
• Competitor Comparison: Your position vs the market
• Action: Should you increase, decrease, or maintain price?

---

🧪 Testing

Run All Tests

# Install dev dependencies
pip install -r requirements-dev.txt

# Run tests
pytest tests/ -v

# Run with coverage
pytest tests/ -v --cov=. --cov-report=html

Test Coverage

Test File	Tests	Coverage
test_utils.py	33	Utility functions, validation
test_model_training.py	18	PricingModel class
test_app.py	24	Validation, elasticity
Total	75	~85%

---

🐳 Docker

Build and Run

# Build image
docker build -t retail-price-optimization .

# Run container
docker run -p 8501:8501 retail-price-optimization

Using Docker Compose

docker-compose up

Access the app at http://localhost:8501.

---

📊 Results & Impact

Metric	Performance	Notes
Model R²	0.62	Solid for noisy retail data
Revenue Uplift	15-20%	On underpriced products
Inference Speed	<0.5s	Real-time capable
Test Coverage	75 tests	All passing

Key Finding

We found products with inelastic demand where customers don't care about small price changes. For these items, we could increase prices significantly (up to 98% revenue boost!) without losing customers.

Real World Impact

• 15-20% more revenue on average (just by pricing smarter)
• Instant recommendations (no more spreadsheet gymnastics)
• Better competitive positioning (know where you stand)
• Clear action items (the model tells you what to do)

---

🔧 Development

Setting Up Dev Environment

pip install -r requirements-dev.txt

Train a New Model

# Basic training
python model_training.py

# With hyperparameter tuning
python model_training.py --tune

Code Quality

# Linting
flake8 .

# Formatting
black .

# Type checking
mypy .

---

🤝 Contributing

1. Fork the repo
2. Create a branch (git checkout -b feature/cool-stuff)
3. Make your changes
4. Run tests (pytest tests/ -v)
5. Push and open a PR

Ideas for Contributions

• Add Neural Network model option
• Build REST API with FastAPI
• Add A/B testing capability
• Improve visualizations
• Add more competitor data sources

---

📄 License

MIT License. Do whatever you want with it.

---

⭐ Star this repo if you find it useful!

🎮 Try the Live Demo