sudrxn/EDA-of-Forest-Fire-Algerian

# 🔥 EDA of Forest Fire in Algeria This project presents an in-depth **Exploratory Data Analysis (EDA)** of forest fire incidents in Algeria, aimed at understanding environmental and meteorological factors that influence wildfire occurrences. Using Python-based tools, this analysis uncovers patterns and correlations that can aid in fire prediction and prevention efforts. 📍 **Dataset Region**: Algeria's forest zones (Bejaia and Sidi Bel-Abbes) 📁 **Source**: UCI Machine Learning Repository - Algerian Forest Fires Dataset --- ## 📌 Objective The primary goal of this EDA is to: - Understand the relationship between meteorological variables and forest fires. - Explore seasonal patterns and fire-prone days. - Visualize and interpret data to extract meaningful insights. --- ## 🧰 Tools & Libraries Used - **Pandas** – for data loading and preprocessing - **NumPy** – for numerical operations - **Matplotlib & Seaborn** – for data visualization - **DateTime & Regex** – for date formatting and parsing --- ## 📊 Key Steps & Insights ✔️ **Data Cleaning** - Combined and structured two separate region datasets - Removed nulls, handled anomalies, and formatted dates ✔️ **Feature Engineering** - Extracted `day`, `month`, and `season` from the date - Created binary classification for `Fire` vs. `No Fire` days ✔️ **Visualization & Analysis** - Heatmaps showing correlation between temperature, wind, humidity, etc. - Bar and line plots comparing fire vs. non-fire days - Region-wise fire trends and seasonal spikes - Observed that **June to September** show maximum fire activity --- ## 📌 Columns in Dataset - **Temperature**, **RH (Relative Humidity)**, **Ws (Wind speed)** - **Rain**, **FFMC**, **DMC**, **DC**, **ISI** (Fire indices) - **BUI**, **FWI** – standard fire weather indexes - **Classes** – Binary flag: `fire` / `not fire` --- ## 📦 How to Run 1. Clone the repository: ```bash git clone github.com cd EDA-of-Forest-Fire-Al …