raunakravi084/Logistic-Regression-for-Algerian-Forest-Fire-Prediction

# Logistic Regression for Algerian Forest Fire Prediction ## Project Overview This project uses **Logistic Regression** to predict the occurrence of forest fires in Algeria based on the Algerian Forest Fires dataset. The project is implemented in a Jupyter notebook (`Logistic Regression using Algerian Forest Fire dataset.ipynb`) using Python and libraries such as scikit-learn, pandas, numpy, seaborn, and matplotlib. The notebook includes data loading, cleaning, exploratory data analysis (EDA), preprocessing, model training, evaluation, and model persistence using pickling. The goal is to classify whether a fire occurred (`fire`) or not (`not fire`) based on meteorological and fire weather index features. The model achieves an accuracy of 96% on the test set, with detailed performance metrics including precision, recall, F1-score, and confusion matrix. ## Dataset The **Algerian Forest Fires dataset** contains 246 records from two regions in Algeria: Bejaia and Sidi Bel-Abbes, collected between June and September 2012. The dataset includes 14 features: - **Date-related features**: day, month, year - **Meteorological features**: Temperature, RH (Relative Humidity), Ws (Wind Speed), Rain - **Fire Weather Index (FWI) components**: FFMC (Fine Fuel Moisture Code), DMC (Duff Moisture Code), DC (Drought Code), ISI (Initial Spread Index), BUI (Buildup Index), FWI (Fire Weather Index) - **Target variable**: Classes (binary: `fire` or `not fire`) - **Derived feature**: region (0 for Bejaia, 1 for Sidi Bel-Abbes) **Source**: - UCI Machine Learning Repository - Kaggle **Note**: The dataset file (`Algerian_forest_fires_dataset_UPDATE.csv`) is required to run the notebook. Download it from the above sources and place it in the project directory. ## Requirements To run the notebook, you need the following Python libraries: - pandas - numpy - seaborn - matplotlib - scikit-learn - statsmodels Install the dependencies using pip: ```bash pip install pandas numpy seaborn matplotl …