SukritDeb/Algerian-Forest-Fire-Predictor

It basically predicts the Fire in the two regions of Algerian Forest # Algerian Forest Fire Predictor This project provides a complete pipeline for analyzing, modeling, and predicting the Forest Fire Weather Index (FWI) in Algeria using machine learning. It includes data cleaning, exploratory data analysis (EDA), model training, and a web application for real-time prediction. ## Features - **Data Cleaning & EDA**: Jupyter notebooks for cleaning and exploring the Algerian forest fire dataset. - **Model Training**: Multiple regression models (Linear, Lasso, Ridge, ElasticNet) with cross-validation and feature engineering. - **Web Application**: Flask-based web app for predicting FWI from user input. ## Dataset - `notebooks/Algerian_forest_fires_dataset_UPDATE.csv`: Raw dataset with meteorological and fire occurrence data for two Algerian regions. - `notebooks/Algerian_forest_fires_dataset_Cleaned.csv`: Cleaned and preprocessed dataset. ## Notebooks - `notebooks/clean-and-eda.ipynb`: Data cleaning, type fixing, region labeling, visualization, and correlation analysis. - `notebooks/model-training.ipynb`: Feature selection, scaling, model training (including saving scaler and model as `.pkl`), and evaluation. ## Web Application - `application.py`: Flask app that loads the trained Ridge regression model and scaler, provides a web interface for FWI prediction. - `templates/index.html`: Welcome page. - `templates/home.html`: Input form for prediction and result display. - `models/ridge_model.pkl`, `models/scaler.pkl`: Saved model and scaler from training notebook. ## Requirements - Python 3.x - Flask - numpy - pandas - scikit-learn Install dependencies with: ```bash pip install -r requirements.txt ``` ## Usage 1. **Data Cleaning & EDA**: Run `notebooks/clean-and-eda.ipynb` to clean and explore the dataset. This generates the cleaned CSV. 2. **Model Training**: Run `notebooks/model-training.ipynb` to train models and export the scaler/model as `.pkl` files. 3. **Web App (Local)**: Start the Flas …