AnushkaSharma05/algerian-forest-fire-prediction

# 🔥 Algerian Forest Fire Risk Prediction A machine learning project that predicts the Fire Weather Index (FWI) for forest fires in Algeria using environmental and fire weather indicators. ## ⚡ Quick Start ```bash pip install -r requirements.txt streamlit run app.py ``` 🌐 **Then open**: localhost ## 🎯 Problem Statement **What Problem?** Forest fires pose significant threats to ecosystems in Algeria. This project predicts the Fire Weather Index (FWI) to enable early detection and risk assessment for better resource allocation and prevention strategies. ## 📊 Dataset - **Source**: Algerian Forest Fires Dataset - **Samples**: ~245 records from 2 regions (Bejaia & Sidi Bel Abbes) - **Features**: Temperature, Humidity, Wind Speed, Rainfall, Fire Indices (FFMC, DMC, DC, ISI, BUI) - **Target**: Fire Weather Index (FWI) ## 🔄 Workflow 1. **EDA** - Data exploration and visualization 2. **Data Cleaning** - Remove missing values and anomalies 3. **Feature Selection** - Remove multicollinear features (correlation > 0.85) 4. **Standardization** - Apply StandardScaler for normalization 5. **Model Training** - Train multiple regression models 6. **Evaluation** - Compare models using MSE, MAE, RMSE, R² ## 🤖 Models Tested | Model | Type | Performance | |-------|------|-------------| | Linear Regression | Baseline | Good | | Lasso Regression | Regularized | Better | | Ridge Regression | Regularized | Better | | ElasticNet | Regularized | Better | | LassoCV | Optimized | **Best** ✓ | | RidgeCV | Optimized | Excellent | | ElasticNetCV | Optimized | Excellent | ## 📈 Results & Performance Metrics ### Model Performance Comparison (Test Set) | Model | MSE | MAE | RMSE | R² Score | |-------|-----|-----|------|----------| | Linear Regression | High | High | High | ~0.50 | | Lasso Regression | Lower | Lower | Lower | ~0.65 | | Ridge Regression | Lower | Lower | Lower | ~0.68 | | ElasticNet | Lower | Lower | Lower | ~0.64 | | **LassoCV** | **Minimal** | **Minimal** | ** …