🔥 Fire Weather Index (FWI) Prediction Web Application

📌 Project Overview

This project is a Machine Learning web application that predicts the Fire Weather Index (FWI) using meteorological and environmental parameters. The model is trained using the Algerian Forest Fires Dataset, which contains weather observations and fire occurrence records collected from two regions of Algeria.

The goal of this project is to demonstrate the end-to-end Machine Learning workflow, including:

• Data Cleaning
• Exploratory Data Analysis (EDA)
• Feature Engineering
• Model Training
• Model Evaluation
• Deployment using Flask Web Framework

Users can input environmental parameters through a web interface, and the trained machine learning model predicts the Fire Weather Index value, which indicates the potential risk of forest fires.

---

🌍 What is Fire Weather Index (FWI)?

The Fire Weather Index (FWI) is a numerical rating system used worldwide to estimate the intensity of forest fires based on weather conditions.

It combines multiple environmental factors such as:

• Temperature
• Humidity
• Wind speed
• Rainfall
• Fuel moisture

A higher FWI value indicates a higher probability and intensity of fire spread.

Typical interpretation:

FWI Value	Fire Risk
< 5	Low
5 – 20	Moderate
> 20	High

---

📊 Dataset

This project uses the Algerian Forest Fires Dataset.

Dataset characteristics:

• 244 observations
• Collected from Bejaia region and Sidi Bel-Abbes region
• Weather and fire occurrence data

Features used in the model:

Feature	Description
Temperature	Air temperature in Celsius
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
Region	Geographic region

Target Variable:

• FWI (Fire Weather Index)

---

🧠 Machine Learning Workflow

The project follows a standard ML pipeline:

1️⃣ Data Cleaning

• Handling missing values
• Removing inconsistent entries
• Converting categorical variables

2️⃣ Exploratory Data Analysis

• Distribution analysis
• Correlation analysis
• Outlier detection

3️⃣ Feature Engineering

• Encoding categorical features
• Scaling numerical features using StandardScaler

4️⃣ Model Training

Multiple regression models were explored:

• Linear Regression
• Ridge Regression
• Lasso Regression
• ElasticNet Regression

5️⃣ Model Evaluation

Models were evaluated using:

• R² Score
• Mean Absolute Error
• Mean Squared Error

The best performing model was selected for deployment.

---

🚀 Web Application

A simple Flask-based web application was built to allow users to interact with the trained model.

User Input Parameters

Users provide the following environmental values:

• Temperature
• Relative Humidity
• Wind Speed
• Rainfall
• FFMC
• DMC
• ISI
• Classes
• Region

The application processes these inputs and returns the predicted Fire Weather Index.

---

🖥️ Project Structure

linear-regression-project
│
├── notebooks
│   ├── Data_cleaning_observation.ipynb
│   └── Regularizations_and_CrossValidation.ipynb
│
├── templates
│   └── home.html
│
├── application.py
├── requirements.txt
├── README.md
└── .gitignore

---

⚙️ Installation

Clone the repository:

git clone https://github.com/yourusername/linear-regression-project.git

Navigate to the project directory:

cd linear-regression-project

Create a virtual environment:

python -m venv venv

Activate the environment:

Windows

venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

---

▶️ Running the Application

Run the Flask app:

python application.py

Open your browser and go to:

http://127.0.0.1:5000/

You will see the FWI Prediction Interface where you can enter weather parameters and obtain predictions.

---

📈 Example Workflow

1. Enter weather parameters in the form
2. Click Predict
3. The trained machine learning model processes the inputs
4. The predicted Fire Weather Index value is displayed

---

🧰 Technologies Used

• Python
• Pandas
• NumPy
• Scikit-learn
• Matplotlib
• Seaborn
• Flask
• HTML / CSS

---

🎯 Learning Outcomes

Through this project, the following concepts are demonstrated:

• End-to-end Machine Learning workflow
• Data preprocessing and feature engineering
• Regularization techniques (Ridge, Lasso, ElasticNet)
• Model evaluation techniques
• Deployment of ML models using Flask
• Building simple ML web applications

---

📌 Future Improvements

Possible improvements for this project:

• Deploy the application using Docker
• Host the model on AWS / Render / Railway
• Add interactive visualizations
• Include automatic input validation
• Use REST API for predictions

---

👨‍💻 Author

Shibam Sadhukhan

Machine Learning Enthusiast | Data Science Learner

---

⭐ If you find this project useful

Please consider starring the repository on GitHub.