nahid-rgb/algerian-forest-fire-prediction
Domain:
environment and energyRecord type:
software
Creator:
nah
Host:
Algerian Forest Fire FWI Prediction using Ridge Regression
# 🔥 Algerian Forest Fire — FWI Prediction
A machine learning web application that predicts the **Fire Weather Index (FWI)** based on weather and environmental conditions in Algeria. Built with Flask and deployed on Render.
🌐 **Live Demo:** algerian-forest-fire-predic…
---
## 📌 What is FWI?
The **Fire Weather Index (FWI)** is a numeric rating of fire intensity used by fire services to assess wildfire danger.
- Higher FWI → More dangerous fire conditions
- Range in this dataset: 0 to 31.1
---
## 📊 About the Dataset
The dataset covers two regions of Algeria and was collected from June to September 2012.
| Property | Details |
|----------|---------|
| Total Records | 244 (122 per region) |
| Regions | Bejaia (northeast) + Sidi-Bel Abbes (northwest) |
| Period | June – September 2012 |
| After cleaning | 243 usable records |
| Fire cases | 137 fire, 106 not fire |
| Target variable | FWI (Fire Weather Index) |
> From EDA: August had the most forest fires in both regions. Most fires occurred across June, July, and August.
---
## 📥 Input Features
The original dataset has 11 attributes. After removing `day`, `month`, `year` (not needed for prediction) and adding `Region`, the model uses these 9 features:
| Feature | Description | Range |
|---------|-------------|-------|
| **Temperature** | Max temperature at noon in Celsius | 22 – 42°C |
| **RH** | Relative Humidity in % | 21 – 90% |
| **Ws** | Wind Speed in km/h | 6 – 29 km/h |
| **Rain** | Total rainfall for the day in mm | 0 – 16.8 mm |
| **FFMC** | Fine Fuel Moisture Code — moisture of fine surface fuels. Higher = drier = higher fire risk | 28.6 – 92.5 |
| **DMC** | Duff Moisture Code — moisture of loosely compacted organic matter | 1.1 – 65.9 |
| **ISI** | Initial Spread Index — expected rate of fire spread | 0 – 18.5 |
| **Classes** | Whether fire occurred: 1 = Fire, 0 = Not Fire | 0 or 1 |
| **Region** | 0 = Bejaia (northeast), 1 = Sidi-Bel Abbes (northwest) | 0 or 1 |
> …