Wildfire Clustering in Europe

Clusterización de incendios forestales en Europa

---

Description

An unsupervised machine learning model that classifies wildfires in Europe (2016–2024) based on location, seasonality, burned area, and vegetation type. Implemented in a reproducible pipeline using EFFIS data and HDBSCAN to identify wildfire patterns and typologies.

Modelo de machine learning no supervisado que clasifica incendios forestales en Europa (2016–2024) según ubicación, estacionalidad, superficie quemada y tipo de vegetación. Implementado en un pipeline reproducible que usa datos de EFFIS y HDBSCAN para identificar patrones y tipologías de incendios.

---

Main Dependencies

• Python 3.10+
• pandas, geopandas
• hdbscan
• scikit-learn
• joblib

---

Installation

1. Clone repository:

   git clone https://github.com/tu_usuario/ML_EFFIS.git
   cd ML_EFFIS
   

2. (Optional) Virtual environment and dependencies::

   python -m venv venv
   source venv/bin/activate  # En Windows: venv\Scripts\activate
   pip install -r requirements.txt
   

Quick Usage

A pretrained version of the model is available at:

/src/models/EFFIS_SCAN.joblib

You can load it directly for testing without running the entire pipeline from scratch:

from joblib import load

# Load pretrained model
EFFIS_SCAN = load("src/models/EFFIS_SCAN.joblib")

# Modify key pipeline parameters with kw_args:
EFFIS_SCAN.set_params(Preprocess_shp__Filter_area__kw_args={'bbox': nuevo_bbox})                # Area of interest
EFFIS_SCAN.set_params(Filter_transform__Feature_select__kw_args={'columns': cols_to_filter})    # Dataset column filter
EFFIS_SCAN.set_params(Filter_transform__Area_filter__kw_args={'ha': new_area})                  # Minimum fire area to consider
EFFIS_SCAN.set_params(Filter_transform__One-Hot__kw_args={'features': new_features_list})       # Vegetation type variables

This allows easy reuse of the pipeline for new geographic areas, custom filters, or analysis configurations.

👉 To run the full project workflow, follow the steps described in main.ipynb (in Spanish).

Credits

Developed by por Víctor Riera
Source: European Forest Fire Information System (EFFIS)