Dharani-kaligi/Algerian-forest-fires-prediction
Domain:
environment and energyclimateRecord type:
dataset
Creator:
Dha
Host:
Machine learning project to predict Fire Weather Index (FWI) using regression models.
# 🔥 Algerian Forest Fires Dataset – EDA, Data Cleaning & Preprocessing
## Project Overview
This project focuses on Exploratory Data Analysis (EDA), data cleaning, feature engineering, and preprocessing of the Algerian Forest Fires dataset. The goal is to understand wildfire patterns and analyze the relationship between meteorological conditions and fire occurrences in two regions of Algeria.
## đź“Ś Overview
Dataset: Algerian Forest Fires (2012)
Instances: 244 (122 from Bejaia, 122 from Sidi Bel-Abbes)
Goal: Analyze forest fire trends, clean the dataset, engineer features, scale data, and explore patterns for future ML modeling.
Tools: Python, Pandas, NumPy, Seaborn, Matplotlib, Scikit-learn
## Objective
To explore how weather and environmental factors influence forest fires and prepare the dataset for machine learning modeling.
## đź“‚ Dataset Information
| Feature | Description |
|---------|-------------|
| Date | From June to September 2012 |
| Temperature | Max temp at noon (°C) |
| RH | Relative Humidity (%) |
| Ws | Wind speed (km/h) |
| Rain | Daily rainfall (mm) |
| FFMC, DMC, DC, ISI, BUI, FWI | Fire Weather Index components |
| Classes | Fire / Not Fire (target variable) |
- 🔢 11 input features + 1 output (`Classes`)
- 🗺️ Two regions: Bejaia (0) and Sidi Bel-Abbes (1)
---
## 🧹 Data Cleaning & Preprocessing
- Removed null values
- Fixed column names and whitespace
- Removed unneeded rows (like row 122)
- Created a new column `Region` based on index
- Converted string/object columns to numerical types
- Encoded target classes (`Fire` → 1, `not fire` → 0)
## ⚖️ Feature Scaling
- Applied scaling to numerical features to normalize the range for model compatibility.
---
đź“Š Exploratory Data Analysis (EDA)
🔍 Techniques Used:
Histograms & Density Plots: Visualize feature distributions and detect skewness.
Boxplots: Identify outliers in numerical features.
Correlation Matrix & Heatmap: Explore relationships between variables.
Pie Chart: Understand the distr …