Shruti-1301/algerian_forestfire_prediction

This project predicts the Fire Weather Index (FWI) and classifies regions into Fire and Not Fire categories using environmental and meteorological data from Algeria. It covers the full machine learning pipeline — from data preprocessing and visualization to model training and evaluation using Linear Regression, Lasso Regression and Ridge Regression # Algerian Forest Fire Prediction using Machine Learning This project predicts the **Fire Weather Index (FWI)** and classifies regions into *Fire* and *Not Fire* categories using environmental and meteorological data from Algeria. It covers the full **machine learning pipeline** — from data preprocessing and visualization to model training and evaluation using **Linear Regression**, **Lasso Regression**, and **Ridge Regression**. ## Dataset Information **Dataset:** Algerian Forest Fires Dataset (UCI Machine Learning Repository) **Total Samples:** 244 (122 from Bejaia region and 122 from Sidi Bel-Abbes region) **Duration:** June 2012 – September 2012 The dataset contains 11 input attributes and 1 output attribute (class). The data represent weather conditions and fire indices recorded during the forest fire season in Algeria. **Features include:** * Date (Day, Month, Year) * Temperature (°C) * Relative Humidity (%) * Wind Speed (km/h) * Rainfall (mm) * Fine Fuel Moisture Code (FFMC) * Duff Moisture Code (DMC) * Drought Code (DC) * Initial Spread Index (ISI) * Buildup Index (BUI) * Fire Weather Index (FWI) * Class (Fire / Not Fire) **Class Distribution:** * Fire: 138 samples * Not Fire: 106 samples ## Project Workflow ### Step 1: Data Preprocessing (`my_code.ipynb`) * Imported the dataset using pandas * Handled missing values and removed unwanted spaces in column names * Converted columns to numeric types where required * Added a new feature “Region” (0 for Bejaia, 1 for Sidi Bel-Abbes) * Encoded the “Classes” column (Fire → 1, Not Fire → 0) * Performed exploratory data analysis (EDA) using matplotlib and seaborn * Created visualizations such as: * Distribution plots for weather attributes * Pie chart showing Fire vs Not Fire ratio * Correlation heatmap for features ### Step 2: Model Development (`my_model_training.ipynb`) * Split the dataset into training and testing sets (75% / 25%) * Standardized the data using **StandardScaler** * Trained three regr …