ManjushaMotamarry/Algerian_Forest_Fire_Prediction

A machine learning project aimed at predicting forest fire severity using meteorological and forest indices, providing actionable insights to mitigate fire risks. Key highlights include data-driven analysis, visualization, and regression models for accurate forecasting. # Forest Fire Prediction: Analyzing and Forecasting Fire Risk ## Table of Contents 1. Introduction 2. Dataset Overview 3. Data Cleaning 4. Exploratory Data Analysis (EDA) 5. Feature Engineering 6. Feature Scaling 7. Results and Insights 8. Conclusion 9. Future Scope --- ## Introduction Forest fires are a major environmental concern, causing extensive damage to biodiversity, property, and human lives. This project leverages machine learning techniques to predict fire severity using meteorological and forest-related indices. The goal is to provide actionable insights that can help forest management teams preemptively allocate resources and mitigate fire risks. --- ## Dataset Overview The dataset comprises **244 instances** of weather data and forest fire observations collected from two Algerian regions: **Bejaia** and **Sidi Bel Abbes**. Key attributes include: - **Weather Features**: Temperature, Relative Humidity, Wind Speed, and Rainfall. - **Fire Weather Index (FWI)**: Indicators like Fine Fuel Moisture Code (FFMC), Duff Moisture Code (DMC), Drought Code (DC), and Initial Spread Index (ISI). - **Outcome**: Binary classes representing fire occurrence (`fire` and `not fire`). --- ## Data Cleaning To ensure data quality, we performed the following steps: 1. **Handling Missing Values**: Removed rows with missing or anomalous data points. 2. **Region Encoding**: Added a new `Region` column for classification (`0` for Bejaia and `1` for Sidi Bel Abbes). 3. **Column Renaming and Type Conversion**: Standardized column names and converted data types for consistency. 4. **Header Removal**: Eliminated redundant headers embedded within the dataset. ### Final Dataset The cleaned dataset includes **243 rows** and **13 columns** with no null values. --- ## Exploratory Data Analysis (EDA) EDA revealed critical patterns in fire occurrence and its relationship with weather conditions: 1. **Fire Distribution**: - **Pie Chart**: A balanced distribution of fire (`56.5%`) a …