AryanKashyap31/Forest_Fire_Algerian

🔥 Algerian Forest Fires Prediction Project This project presents a full-stack, end-to-end machine learning pipeline deployed as a web app to predict fire risk in Algerian forests. The system uses real meteorological data and the Canadian Forest Fire Weather Index System features to estimate fire occurrence, making it a practical tool for environmental monitoring and resource management. Project Structure 1. Data Source: The core data is from Algerian_forest_fires_dataset_UPDATE.csv, covering two regions (Bejaia & Sidi-Bel Abbes), with weather and fire indices and a target variable (Classes: fire/not fire). 2. Data Cleaning & Preparation: Notebook: dataCleaning.ipynb Loads, cleans, merges, and preprocesses the raw dataset. Handles region split, removes headers and footers, drops nulls and inconsistent entries. Converts categorical labels ("fire"/"not fire") and region info into numeric codes for modeling. Performs exploratory data analysis (EDA), including correlation analysis. Output: Algerian_forest_fires_cleaned_dataset.csv. 3. Model Training: Notebook: model_Training.ipynb Loads the clean data. Trains a machine learning classifier (Ridge Regression, possibly for FWI or binary classification?) to predict fire occurrence, using features like Temperature, Humidity, Wind Speed, ISI, etc. Scales features, evaluates performance, and saves the model and scaler as pickled objects for deployment (models/ridge.pkl, models/scaler.pkl). 4. Web Application: Backend: application.py Flask-based backend which loads the model and scaler. Exposes two routes: / : Home/index page. /predictdata : Form-based prediction endpoint. Accepts input features from web form, applies standard scaling, and outputs prediction (renders result to home.html). Frontend: index.html, home.html (templates with input forms and results display). 5. Requirements: Dependencies listed in requirements.txt: Flask, numpy, pandas, scikit-learn, gunicorn. Running Locally Install dependenc …