GururajIngavale/FireSense-Algeria

This project highlights how machine learning can be leveraged to predict and mitigate forest fire risks by analyzing weather data. The substantial improvement in model accuracy underscores the value of feature selection, regularization techniques (Ridge/Lasso), and ensemble methods in building robust, real-world predictive systems. 🔥 Algerian Forest Fire Prediction Using Machine Learning (FireSense Algeria) --- ## 🌍 Overview Forest fires are one of the most devastating natural disasters, causing environmental destruction, loss of biodiversity, and economic damage. This project leverages **Machine Learning** to predict the likelihood of forest fires based on weather conditions in the **Bejaia** and **Sidi Bel-Abbès** regions of Algeria (June–September 2012). --- ## 📊 Dataset - **Source:** Algerian Forest Fires Dataset - **Size:** 244 weather records - **Features:** 11 weather & fire indices (Temperature, Relative Humidity, Wind Speed, Rainfall, Fire Weather Index components, etc.) - **Target Variable:** - 🔥 Fire (138 instances) - 🌱 No Fire (106 instances) --- ## 🎯 Objective Predict the **probability of a forest fire** using multiple machine learning algorithms and compare their performance. --- ## ⚙️ Methodology ### 🧹 Data Preprocessing ✔️ Handled missing values ✔️ Normalized numerical attributes ✔️ Encoded categorical variables (dates) ✔️ Conducted Exploratory Data Analysis (EDA) ### 🤖 Models Implemented - Logistic Regression - Ridge & Lasso Regression - Decision Trees / Random Forests - Support Vector Machine (SVM) - Gradient Boosting ### 🔧 Model Optimization - Hyperparameter tuning with **GridSearchCV** - Applied **Cross-Validation** - Compared metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC --- ## 📈 Results | Model | Accuracy | |---------------------------|-----------| | Logistic Regression | ~70% | | Ridge / Lasso Regression | Improved | | Random Forest / Gradient Boosting | **85–90%** ✅ | 🔥 **Key Predictors:** Temperature, Relative Humidity, Wind Speed, and FWI indices --- ## 🚀 Getting Started ### 1️⃣ Clone the Repository ```bash git clone github.com cd algerian-forest-fire-prediction