harshil2937/Algerian-Forest-Fire-Prediction

It is a Beginner Machine Learning project using scikit-learn and FlaskAPI for predicting Fire Whether Index. # Algerian-Forest-Fire-Prediction It is a Beginner Machine Learning project using scikit-learn and FlaskAPI for predicting Fire Weather Index. ## Project overview Forest fires cause significant environmental and economic damage every year. This project aims to predict the "Fire weather index(FWI)" based on meteorological conditions such as temperature, humidity, wind speed, rainfall, and fire weather indices. The trained machine learning model is integrated into a Flask web application, allowing users to enter weather parameters and receive predictions through a simple web interface. ## Dataset The dataset contains weather observations collected from two regions in Algeria. - Temperature: Current air temperature in °C. - Relative Humidity (RH): Current relative humidity as a percentage 0-100%. - Wind Speed (Ws): Current wind speed in km/h. - Rain: Amount of rainfall in millimeters (mm). - FFMC: Fine Fuel Moisture Code indicating the moisture content of surface fuels. - DMC: Duff Moisture Code indicating the moisture content of the middle organic fuel layer. - DC: Drought Code indicating the long-term dryness of deep organic fuels. - ISI: Initial Spread Index indicating the expected rate at which a fire may spread. - BUI: Build-Up Index indicating the total amount of combustible fuel available. Target variable : Fire weather index(FWI). ## Machine learning models The following regression algorithms were trained and compared: - Linear Regression - Ridge Regression - RidgeCV - Lasso Regression - LassoCV - ElasticNet - ElasticNetCV The best-performing model was saved and deployed using Flask. ## Future Improvements - Perform hyperparameter tuning using GridSearchCV - Apply Cross Validation - Add more visualizations for EDA - Deploy the application on Render or Railway - Improve the web interface with Bootstrap - Implement model monitoring and logging