Mehul-kh2005/Algerian-Forest-Fires-Analysis

🔥 End-to-end regression project on the Algerian Forest Fires dataset to predict FWI using ML models (Linear, Ridge, Lasso, ElasticNet). Includes EDA, preprocessing, model training & evaluation, model saving (pickle), and a Flask web app for real-time prediction. # Algerian Forest Fires Prediction Analysis ## Project Overview 1. This project analyzes the Algerian Forest Fires dataset and aims to predict the fire weather index (FWI) using various regression models. 2. The models used in this project include Linear Regression, Lasso, Ridge, ElasticNet, and their cross-validated versions. 3. The dataset contains information about fire incidents across different regions of Algeria, with various features influencing fire behavior. 4. A **Flask web application** has been developed for real-time prediction using the best-trained regression model. ### Key Features: - **Data Preprocessing**: Cleaning and feature engineering, including handling missing values, feature scaling, and encoding categorical variables. - **Exploratory Data Analysis (EDA)**: Visualizing the dataset to find patterns and correlations. - **Model Building**: Applying multiple regression techniques to predict the FWI. - **Hyperparameter Tuning**: Using cross-validation to fine-tune model parameters for better performance. - **Performance Evaluation**: Evaluating the models using Mean Absolute Error (MAE) and R² score. - **Flask Web App:** A lightweight web app using Flask for interactive FWI prediction based on user input. - **Model Deployment:** Model serialization using `pickle` for deployment in the Flask app. ## 📂 File Structure The repository contains the following files: - **Algerian_forest_fires_dataset.ipynb**: Data cleaning and exploratory data analysis (EDA) phase. - **Model_Training.ipynb**: Building and evaluating multiple regression models. - **application.py:** Main Flask app for serving the prediction web interface. - **templates/:** HTML templates for the web interface. - **models/:** Contains the saved trained model (Pickle file) used for deployment. - **images/**: Folder containing visualizations generated during the analysis (e.g., scatter plots, box plots, and heatmaps). ```bash Algerian-Forest-Fires-Analysis/ ├── data/ │ ├── Algerian_ …