yahia-zakaria/algerian-forest-fire-fwi-prediction

# 🔥 Algerian Forest Fire Intensity Prediction using FWI (Lasso Regression) A regression-based machine learning project that predicts the **Fire Weather Index (FWI)** — a key indicator of forest fire danger — in Algeria. The project uses **Lasso Regression** to model fire intensity based on meteorological and environmental conditions. --- ## 📂 Project Overview Algeria, with its Mediterranean climate, is vulnerable to seasonal forest fires. This project aims to: - Predict the **FWI (Fire Weather Index)** using key weather variables - Apply **Lasso Regression** for both modeling and feature selection - Evaluate performance using **R² Score** and **Mean Absolute Error (MAE)** - Visualize the most influential predictors of fire risk --- ## 📊 Dataset - **Source**: UCI Algerian Forest Fires Dataset - **Target Variable**: `FWI` (Fire Weather Index) - **Input Features**: - Temperature (°C) - Relative Humidity (%) - Wind (km/h) - Rain (mm) - FFMC, DMC, DC, ISI, BUI (fire indices) - Day, Month --- ## 🧠 Machine Learning Approach - **Model Used**: Lasso Regression (`L1`-penalized linear regression) - **Why Lasso?** - Automatically reduces irrelevant features - Improves generalization by shrinking coefficients - Especially useful when dealing with multicollinearity --- ## 📈 Performance Metrics | Metric | Value | |------------------------|-----------| | R² Score | 0.98 | | Mean Absolute Error | 0.620 | > ✅ A high R² score indicates that the model explains most of the variance in the target FWI. > ✅ Low MAE indicates accurate and stable predictions. --- ## 📌 Project Structure