rushee-67/Algerian-Forest-Fires-ML

# 🔥 Algerian Forest Fires Prediction – End-to-End ML Application 🚀 **Live Demo:** algerian-forest-fires-ml.on… ## 📌 Problem Statement Forest fires cause significant environmental and economic damage. This project aims to **predict forest fire severity indicators** using meteorological and environmental features from the **Algerian Forest Fires dataset**. The project demonstrates a **complete applied machine learning workflow**: data analysis → model training → model persistence → web-based inference. --- ## 📊 Dataset - **Source:** Algerian Forest Fires Dataset - **Description:** Weather and fire-related environmental attributes - **Target:** Fire severity–related continuous values - **Location:** `data/` directory --- ## 🔍 Exploratory Data Analysis & Feature Engineering 📘 **Notebook:** `notebooks/eda_feature_engineering.ipynb` ### Key Steps - Data cleaning and handling inconsistent values - Exploratory Data Analysis (EDA) to analyze: - Feature distributions - Outliers and anomalies - Relationships between meteorological variables - Correlation analysis to identify **multicollinearity** - Feature transformation and preparation for regression models - Scaling numerical features using **StandardScaler** ### Key Insights - Meteorological features such as temperature, wind speed, and humidity show meaningful influence on fire severity - Presence of multicollinearity motivated the use of **regularized regression models** - Feature scaling is essential for stable model training --- ## 🤖 Model Training & Evaluation 📕 **Notebook:** `notebooks/model_training.ipynb` ### Models Implemented - Linear Regression (baseline) - Lasso Regression (with cross-validation) - Ridge Regression - Elastic Net Regression ### Model Selection Regularized regression models were explored to handle multicollinearity and improve generalization. **Ridge Regression** was selected as the final model due to its stable performance and balanced bias–variance trade-off. ### …