🔥 Algerian Forest Fire Weather Index Prediction

Deployed on Render Link :https://fire-whether-index-prediction.onrender.com/

📌 Overview

This project focuses on predicting the Fire Weather Index (FWI) using regression-based machine learning models. FWI is a continuous indicator that represents the potential intensity and severity of forest fires.

The model is trained on the Algerian Forest Fires dataset, which contains meteorological and fire-related measurements from two regions of Algeria.

---

🎯 Problem Statement

Forest fires cause significant environmental and economic damage. Accurately estimating fire severity in advance is crucial for prevention and disaster management.

Goal:
Build a machine learning regression model that predicts the Fire Weather Index (FWI) based on weather conditions, fire indicators, and regional information.

---

🌍 Dataset Description

The dataset consists of daily observations collected during the high fire-risk season (June–September).

🔹 Regions Covered

• Bejaia Region (Region = 0)
• Sidi-Bel-Abbes Region (Region = 1)

---

📊 Input Features

• Temperature – Ambient temperature (°C)
• RH – Relative Humidity (%)
• Ws – Wind speed (km/h)
• Rain – Rainfall amount (mm)
• FFMC – Fine Fuel Moisture Code
• DMC – Duff Moisture Code
• ISI – Initial Spread Index
• Classes – Fire occurrence indicator
  • 0 → No Fire
  • 1 → Fire
• Region – Geographical region
  • 0 → Bejaia
  • 1 → Sidi-Bel-Abbes

🎯 Target Variable

• FWI (Fire Weather Index) – Continuous value indicating fire severity

---

🧠 Machine Learning Approach

• Problem Type: Regression
• Pipeline Steps:
  • Data cleaning and preprocessing
  • Feature scaling
  • Handling regional and class indicators
  • Training multiple regression models
  • Selecting the best-performing model

---

📈 Model Evaluation

The regression model is evaluated using:

• R² Score
• Mean Absolute Error (MAE)
• Mean Squared Error (MSE)
• Root Mean Squared Error (RMSE)

---

🔬 Experiment Tracking

• MLflow is used for:
  • Logging model parameters
  • Tracking evaluation metrics
  • Saving trained models
• Experiments are tracked using DAGsHub MLflow UI

---

🌐 Web Application

A Flask-based web application allows users to:

• Enter meteorological and fire-related inputs
• Predict Fire Weather Index (FWI) instantly
• View results in a responsive UI

---

🚀 Deployment

• Deployed on Render
• Production-ready setup using Gunicorn
• Linux-compatible dependencies

---

🛠 Tech Stack

• Language: Python
• Libraries: NumPy, Pandas, Scikit-learn
• Experiment Tracking: MLflow, DAGsHub
• Web Framework: Flask
• Deployment: Render
• Version Control: Git, GitHub

---

▶️ How to Run Locally

1️⃣ Clone the repository

git clone https://github.com/your-username/Algerian-Forest-Fire-Prediction.git
cd Algerian-Forest-Fire-Prediction

python -m venv venv
venv\Scripts\activate   # Windows

pip install -r requirements.txt

python app.py

run above line in step by step on CMD

open in browser -- http://127.0.0.1:5000