How I Built a Dynamic Pricing ML System

Introduction

A Dynamic Pricing ML System that predicts optimal prices for retail products to maximize sales revenue. Built with LightGBM, Flask, and a beautiful web UI.

🎯 What Does This Project Do?

Given a product (identified by stockcode), this system predicts:

How many units will sell at different price points
What is the optimal price to maximize total revenue

Business Question:

"If I price product X at $Y, how many will I sell and what's my revenue?"

✨ Key Features

Feature	Description
🤖 ML Models	LightGBM + ElasticNet ensemble for robust predictions
🌐 REST API	Flask-based API with CORS support
🎨 Web UI	Beautiful, responsive prediction interface
📊 Visualizations	Interactive charts showing price vs. sales curves
🚀 Production Ready	Docker support for AWS Lambda/SageMaker deployment
⚡ Fast Inference	~1-2 seconds per prediction

🏗️ System Architecture

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   Raw Data      │────→│  Data Pipeline  │────→│  Processed Data │
│  (CSV/Excel)    │     │  (Engineering)  │     │   (Parquet)     │
└─────────────────┘     └─────────────────┘     └─────────────────┘
                                                        │
                                                        ↓
┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   Predictions   │←────│   Flask API     │←────│  ML Models      │
│   (JSON/UI)     │     │   (Waitress)    │     │  (LightGBM)     │
└─────────────────┘     └─────────────────┘     └─────────────────┘

Quick Start

bash

# Clone the repository
git clone https://github.com/danishsyed-dev/ml-sales-prediction.git
cd ml-sales-prediction

# Install dependencies
pip install -r requirements.txt

# Run the application
python application.py