diptobarua88/Forest-Fire-Prediction

🔥 Forest Fire Prediction Developed a machine learning project using the Algerian Forest Fires Processed Dataset, applying Random Forest and Logistic Regression models to predict fire occurrences. The project included EDA, feature engineering, correlation heatmap, SMOTE balancing, PCA, and SHAP-based explainability for model interpretation. # 🔥 Forest Fire Prediction This project focuses on predicting forest fire occurrences using machine learning techniques. Built with the **Algerian Forest Fires Processed Dataset**, it applies both **Random Forest** and **Logistic Regression** models to classify and understand fire behavior. --- ## 📊 Project Overview The goal of this project is to predict whether a forest fire will occur based on meteorological and environmental factors. Comprehensive data analysis and advanced ML techniques were used to improve accuracy and interpretability. --- ## 🧠 Key Features - **Exploratory Data Analysis (EDA):** Visualized dataset distributions, outliers, and trends to understand fire-related patterns. - **Feature Engineering:** Cleaned and transformed features for optimal model performance. - **Correlation Heatmap:** Identified strong relationships between variables. - **Data Balancing (SMOTE):** Handled class imbalance using Synthetic Minority Over-sampling Technique. - **Dimensionality Reduction (PCA):** Reduced feature space while preserving important variance. - **Model Building:** Implemented and compared: - 🌲 **Random Forest Classifier** - 📈 **Logistic Regression** - **Explainability (SHAP):** Used SHAP values for interpretable AI, explaining feature importance and individual predictions. --- ## ⚙️ Tech Stack - **Language:** Python - **Libraries:** `pandas`, `numpy`, `matplotlib`, `seaborn`, `scikit-learn`, `imblearn`, `shap` --- ## 📁 Project Structure