Sohaib4238/Algerian_Forest_Project

# Algerian Forest Fire Predictor An end-to-end Machine Learning project that predicts the **Fire Weather Index (FWI)** based on Algerian forest weather data. This solution handles multicollinearity using Ridge and Lasso regression and is deployed as a Flask web application. ## Project Overview * **Goal:** Predict forest fire risks (FWI) using environmental features like Temperature, Humidity, Rain, and Wind Speed. * **Accuracy:** Achieved **98.7% R² score** using Ridge Regression. * **Techniques:** Solved high multicollinearity between features (DC, BUI, DMC) using Regularization (Lasso/Ridge). * **Deployment:** Built a responsive web interface with **Flask** and deployed on **AWS EC2**. ## Tech Stack * **Language:** Python 3.8+ * **Libraries:** Scikit-learn, Pandas, NumPy, Seaborn, Matplotlib * **Web Framework:** Flask * **Cloud:** AWS EC2 (Elastic Compute Cloud) ## Model Performance I trained and compared three models to find the best fit: | Model | R² Score | Description | | :--- | :--- | :--- | | **Linear Regression** | 98.9% | Baseline model (prone to overfitting due to multicollinearity). | | **Lasso Regression** | 98.4% | Used L1 Regularization to feature selection. | | **Ridge Regression** | **98.7%** | **(Selected Model)** Best balance of bias-variance trade-off. | ## How to Run Locally 1. **Clone the repository:** ```bash git clone github.com cd Algerian_Forest_Project ``` 2. **Install dependencies:** ```bash pip install -r requirements.txt ``` 3. **Run the Flask App:** ```bash python application.py ``` 4. **Access the App:** Open your browser and go to `127.0.0.1` ##Project Structure ```text Algerian_Forest_Project/ ├── dataset/ # Raw and Cleaned Data ├── models/ # Pickled Models (ridge.pkl, scaler.pkl) ├── templates/ # HTML Frontend (index.html, home.html) ├── application.py # Flask Main Application ├── training.ipynb # …