HrideshMaithani14/Algerian-Forest-Fire-Prediction

# Algerian Forest Fire FWI Prediction App A **real-time Fire Weather Index (FWI) prediction web application** built with **Flask**, **Scikit-learn**, and **Ridge Regression**. Uses the **Algerian Forest Fires Dataset** to predict fire danger based on weather and fuel conditions. --- ## Features - **Real-time FWI Prediction** via interactive web form - **Side-by-side layout**: Input on left, result with **animated fire** on right - **Fire Status & Region** dropdowns (Bejaia / Sidi Bel-abbes) - **ISI field highlighted** for emphasis - **Responsive & Mobile-Friendly** - **Glassmorphism UI** with floating fire sparks - **No rounding** — exact model output - **Trained on 244 real fire events (2012)** --- ## Dataset - **Source**: Algerian Forest Fires Dataset - **Regions**: Bejaia (0), Sidi Bel-abbes (1) - **Features Used**: - `Temperature`, `RH`, `Ws`, `Rain` - `FFMC`, `DMC`, `ISI` - `Classes` (fire / not fire) - `Region` --- ## Model Performance | Model | R² Score | MAE | |---------------|----------|------| | **Ridge Regression** | **0.98** | **0.45** | | Lasso | 0.96 | 0.72 | | Linear | 0.97 | 0.58 | **Ridge selected** for best balance of accuracy & robustness. --- ## Tech Stack | Technology | Purpose | |------------------|--------| | **Python** | Core logic | | **Flask** | Web framework | | **Scikit-learn** | Model training & scaling | | **Pandas/NumPy** | Data processing | | **HTML/CSS/JS** | Frontend (glassmorphism + animations) | | **Bootstrap 5** | Responsive design | | **Pickle** | Model persistence | --- ## Project Structure Algerian-Forest-Fire-Prediction/ │ ├── application.py # Flask app ├── model/ │ ├── ridge.pkl # Trained model │ └── Scaler.pkl # StandardScaler ├── templates/ │ ├── index.html # Landing page │ └── home.html …