shivrajsingh08/Algerian_Forest_Fire_Regression_Model

# 🔥 Algerian Forest Fire Classification - ML Mini Project This project builds a machine learning model to classify the presence of forest fires in Algerian regions using meteorological data. It covers the full ML pipeline: from data cleaning to feature engineering, modeling, evaluation, and model deployment using Pickle. --- ## 🧠 Objective Predict the FWI(Fire Weather Index) based on environmental conditions like temperature, humidity, wind speed, and drought indices. --- ## 📊 Dataset - *Source:* Algerian Forest Fires Dataset (UCI ML Repository) - *File Used:* Algerian_forest_fires_dataset_UPDATE.csv - *Total Samples:* 247 - *Features:* 12 (e.g., Temperature, RH, WS, FFMC, DMC, DC, ISI, BUI, FWI) - *Target Variable:* FWI --- ## ⚙ Project Workflow 1. *Data Loading & Cleaning* - Remove redundant header rows - Handle missing values 2. *Exploratory Data Analysis* - Correlation heatmaps - Class distribution - Feature histograms 3. *Feature Engineering* - Label Encoding for target variable - Standardization using StandardScaler 4. *Model Training* - Logistic Regression - Lasso Regression - Ridge Regression - Elastic Net Regression - Train-test split (typically 80/20) 5. *Evaluation* - R2 Score - Erroe Matrices 6. *Model Deployment* - Save model and scaler using pickle - model.pkl and scaler.pkl files included --- ## 💻 Project Structure ``` Algerian-Forest-Fire-ML/ ├── algerian_forest_fire_model_prediction.ipynb → Main notebook: EDA, feature engineering, modeling ├── Algerian_forest_fires_dataset_UPDATE.csv → Raw dataset (original source) ├── Algerian_forest_fires_model_cleaned_dataset.csv → Cleaned dataset used for training ├── model.pkl → Pickled trained model ├── scaler.pkl → Pickled scaler object ├── requirements.txt → Python dependencies └── README.md …