akshat310/AlgerianForestFire-Linear-Ridge-Lasso-ElasticNet-
Domain:
environment and energyRecord type:
project
Creator:
aks
Host:
This project predicts the occurrence of forest fires in Algeria based on meteorological and environmental data.
# 🌲 Algerian Forest Fire Prediction 🔥
## 📌 Project Overview
This project predicts the likelihood of **forest fires in Algeria** using environmental and meteorological parameters.
We apply multiple **machine learning regression models** — **Linear Regression**, **Ridge**, **Lasso**, and **Elastic Net** — to historical fire data to identify key factors influencing fire occurrences.
---
## 📂 Dataset Description
**Name:** Algerian Forest Fires Dataset *(Cleaned)*
**Regions Covered:**
- 📍 **Bejaia** – North-East Algeria
- 📍 **Sidi Bel-abbes** – North-West Algeria
### Features in Dataset
| Feature | Description |
|---------|-------------|
| **Temperature** | Measured in °C |
| **RH** | Relative Humidity (%) |
| **Wind** | Wind Speed (km/h) |
| **Rain** | Rainfall (mm/m²) |
| **FFMC** | Fine Fuel Moisture Code |
| **DMC** | Duff Moisture Code |
| **DC** | Drought Code |
| **ISI** | Initial Spread Index |
| **BUI** | Buildup Index |
| **FWI** | Fire Weather Index |
| **Classes** | `fire` / `not fire` |
---
## ⚙️ Technologies Used
- 🐍 **Python**
- 📊 **Pandas**, **NumPy** – Data Processing
- 📈 **Matplotlib**, **Seaborn** – Data Visualization
- 🤖 **Scikit-learn** – Machine Learning Models
---
## 🔍 Methodology
### 1️⃣ Data Preprocessing
- Handle missing values
- Encode categorical features (`Classes`)
- Standard scaling of features
- Remove high correlation to avoid **multicollinearity**
### 2️⃣ Exploratory Data Analysis (EDA)
- Feature distributions for `fire` vs `not fire`
- Correlation heatmaps & feature importance
### 3️⃣ Model Building
- **Linear Regression**
- **Ridge Regression**
- **Lasso Regression**
- **Elastic Net Regression**
### 4️⃣ Model Evaluation
- **R² Score**
- **Mean Squared Error (MSE)**
- **Mean Absolute Error (MAE)**
---
## 📊 Results
- **Comparison of regression models** to determine the most accurate for predicting forest fire likelihood.
- **Insights into the parameters** most strongly correlated with fire occurrences.
---