ShreyanshKun/algerian_forest_fires

# Algerian Forest Fires Prediction 🌲🔥 ## 📌 Project Overview This project focuses on predicting the Fire Weather Index (FWI) and understanding the key factors contributing to forest fires in Algeria. Using the **Algerian Forest Fires Dataset**, the project walks through a complete end-to-end Machine Learning lifecycle, including Data Cleaning, Exploratory Data Analysis (EDA), Feature Engineering, and Model Training. The predictive modeling is handled using multiple regression algorithms to find the best fit for the data: * Multiple Linear Regression * Ridge Regression * Lasso Regression * Elastic Net Regression ## 📊 Dataset Information The dataset includes instances from two regions of Algeria: 1. **Bejaia Region** (Northeast of Algeria) 2. **Sidi Bel-abbes Region** (Northwest of Algeria) It contains weather data and fire-related indices collected over several months. Key features include Temperature, Relative Humidity (RH), Wind Speed (Ws), Rain, and various components of the FWI system. ## 🛠️ Project Lifecycle & Workflow 1. **Data Cleaning:** Handled missing values, removed inconsistent string characters, and converted data types to proper numeric formats. 2. **EDA & Feature Engineering:** Visualized data distributions using Matplotlib and Seaborn to understand correlations and outliers. Handled categorical features. 3. **Feature Scaling:** Applied `StandardScaler` to ensure all numerical inputs are on the same scale for the regression models. 4. **Model Training:** Trained multiple regression models and evaluated them using metrics like R-squared (R²) and Mean Absolute Error (MAE). 5. **Model Pickling:** Saved the best-performing model (`ridge.pkl`) and the standard scaler (`scaler.pkl`) for future deployment. ## 🗂️ Repository Structure * `2.0-EDA And FE Algerian Forest...ipynb`: Notebook containing Data Cleaning, EDA, and Feature Engineering. * `3.0-Model Training.ipynb`: Notebook covering the training, testing, and evaluation of the regression models. * `m …