Bridging the gap between raw transactional data and predictive intelligence. Smart Retail Predictor is a full-stack, AI-driven inventory management and sales forecasting platform. It seamlessly ingests raw Excel data, processes it through a highly normalized MySQL database engine, and utilizes Machine Learning to generate 30-day demand forecasts—all visualized on a sleek, modern glassmorphism web application.
- Demand Forecasting: Utilizes Machine Learning (
scikit-learn) to project 30-day sales trends, factoring in historical velocity and seasonal indexing. - Automated Data Pipelines: Drag-and-drop file ingestion system that automatically cleanses, normalizes, and maps bulk CSV/Excel data to the relational backend.
The backbone of the application is a Fully Normalized (3NF) MySQL Schema. The system offloads heavy computational logic directly to the database layer for maximum performance.
- Complex Query Execution: Utilizes advanced SQL techniques including Correlated Subqueries, Set Operations (
UNION), and Advanced Aggregations (HAVING) to generate instant reports. - Transaction Integrity: Strict error handling prevents bad data injection, with automated logging systems tracking bulk upload actions.
Recently migrated from a monolithic Streamlit architecture to a modern, decoupled client-server model:
- FastAPI Backend: Provides robust, asynchronous RESTful API endpoints for data processing and machine learning integration.
- Custom Glassmorphism Frontend: A highly responsive, premium dark-themed UI built with pure HTML, CSS, and vanilla JavaScript.
- Secure Authentication: Seamless username and password verification with secure tokens.
- Python 3.9+
- MySQL Server (running locally)
We've included simple scripts to manage the application server seamlessly:
- Start the Server: Double-click
start_server.bat - Access the App: Open your browser and navigate to
http://localhost:8000 - Stop the Server: Double-click
stop_server.batto instantly kill all related background processes.
The custom web UI provides a unified command center for store managers:
- Real-Time KPIs: Instant tracking of revenue, quantity sold, and user activity.
- Data Ingestion: Secure drag-and-drop file uploads for continuous data integration.
- Visual Data Storytelling: Interactive charts and data frames mapping historical performance against AI-generated future predictions.
| Category | Technologies Used |
|---|---|
| Frontend UI | HTML5, CSS3 (Glassmorphism), Vanilla JavaScript, Chart.js |
| Backend API | Python, FastAPI, Uvicorn |
| Database | MySQL 8.0 |
| Machine Learning | Scikit-learn, NumPy, Pandas |
| Authentication | Secure Local Auth |
| Version Control | Git, GitHub |