Demand Forecasting App

Predict product demand with ensemble ML, confidence intervals, and an interactive dashboard.

Overview

This Streamlit app helps teams forecast demand using an ensemble of models (XGBoost, LightGBM, Random Forest). It supports 1–12 month horizons, shows confidence bands, and keeps data fresh with a scheduled daily run.

Key Features

Smart ensemble predictions with interactive charts & insights.
Data freshness indicator and system monitoring panel.
Auto-processing of new data daily at 02:00.
Deploy via Docker or run in a Python env; includes demo data to try quickly.

What You’ll See

Live Dashboard: real-time metrics, top products, trends, freshness.
Forecasting: per-product forecasts, 1–12 months, CSV export.
Monitoring: job status, manual refresh, health checks.

Quick Start

Clone the repo and run docker-compose up -d, then open http://localhost:8501.
To use your data, provide a CSV with product_id, date, sales_quantity, unit_price and set the path in .env.

Repository

GitHub – Demand Forecasting App