Weather-Forecasting-AI

🎯 Project Overview This project tackles a critical meteorological challenge: Predicting local rainfall with high precision. unlike standard forecasts, this system uses a 2-Stage Stacked Pipeline to answer two questions: Classification: Will it rain tomorrow? (Yes/No) Regression: If yes, how much rain will fall? (mm) The model was trained on 2,000+ observations containing complex atmospheric data, utilizing advanced feature engineering to capture non-linear weather patterns. The system is built on a dual-model architecture: A[Input Data] --> B{Classifier Model} B -- No Rain --> C[Result: Dry Day ☀️] B -- Rain Expected --> D[Regressor Model] D --> E[Result: Rain Intensity (mm) 🌧️] 1. Classification Stage (Logistic Regression) Goal: Filter out dry days to reduce noise. Technique: Balanced class weights to handle data imbalance. 2. Regression Stage (Random Forest) Goal: Estimate rainfall quantity for positive cases. Configuration: 100 Trees, Max Depth 15, optimized for minimal RMSE.