HaZaRdOuSDeVeLoPeR/Algerian-Forest-Fire-Predictor

Predict the chance of Forest Fires using some input parameters # 🌲🔥 Algerian Forest Fire Predictor A Machine Learning web application built with **Flask** that predicts **Fire Weather Index (FWI)** and **forest fire occurrence (Fire / No Fire)** for Algerian forests using meteorological and vegetation indices. The app uses pre-trained regression and classification models to provide real-time fire risk insights. --- ## 📘 Overview This project uses the **Algerian Forest Fires Dataset** to model and predict: - **Fire Weather Index (FWI)** using regularized linear regression models. - **Fire Occurrence (Fire / No Fire)** using logistic regression classifiers. Users provide meteorological inputs through a web interface, and the application returns both the predicted FWI value and the likelihood of a forest fire. --- ## 🚀 Features ### 🔥 Fire Risk Prediction (Classification) Predicts whether there is a **risk of forest fire** using Logistic Regression-based models. - GridSearchCV-optimized Logistic Regression - RandomizedSearchCV-optimized Logistic Regression These models classify the input as **Fire** or **No Fire**, based on learned decision boundaries from the dataset. ### 📈 FWI Prediction (Regression) Predicts the **Fire Weather Index (FWI)**, an indicator of potential fire intensity. Models used with cross-validated hyperparameter tuning: - **Ridge Regression** (L2 regularization) - **Lasso Regression** (L1 regularization) - **ElasticNet Regression** (combined L1 + L2) Each model is trained on scaled features and saved as a `.pkl` file for fast inference. ### 🖥️ Web Application - Built using **Flask** - User-friendly **HTML + CSS** interface - Takes meteorological inputs from an HTML form - Uses serialized `.pkl` models for live predictions - Displays: - Fire risk: **Fire / No Fire** - Predicted **FWI** value --- ## 🧠 Machine Learning Models ### 🔷 FWI Regression Models | Model | Technique | Notes | |--------------------|-------------------|--------------------------- …