kratikgupta77/Algerian-forest-fire-ml-predict

# Algerian-forest-fire-ml-predict ## Algerian Forest Fire Prediction using Machine Learning This repository contains an end-to-end pipeline for Exploratory Data Analysis (EDA) and the implementation of multiple machine learning models to predict forest fires in Algeria. The primary goal is to use historical weather and fire data to identify fire-prone days and improve early warning systems. ## About the Dataset The dataset is collected from two regions in Algeria — Bejaia and Sidi Bel-Abbes — covering the period from June to September 2012. It includes meteorological observations and Fire Weather Index (FWI) components. ## Features include: Date (DD/MM/YYYY) Weather Data: Temp – Temperature at noon (°C) RH – Relative Humidity (%) Ws – Wind Speed (km/h) Rain – Total daily rain (mm) FWI System Components: FFMC – Fine Fuel Moisture Code DMC – Duff Moisture Code DC – Drought Code ISI – Initial Spread Index BUI – Buildup Index FWI – Fire Weather Index ## Classes: fire: Day with fire occurrence not fire: Day without fire ## Exploratory Data Analysis (EDA) The dataset is cleaned and visualized to reveal trends, correlations, and patterns associated with forest fire occurrences. Key EDA tasks include: Handling missing values Visualizing regional distributions Feature correlation analysis Class distribution and imbalance handling ## Machine Learning Models Used The following supervised ML classification algorithms were trained and evaluated: Logistic Regression Decision Tree Classifier Random Forest Classifier XGBoost Classifier Each model was tuned and evaluated based on accuracy, precision, recall, and F1-score. Results The models were compared to identify the most effective approach for predicting forest fires. XGBoost and Random Forest performed particularly well in balancing bias and variance.