geoRF: a Geospatial Random Forest 🌳

GeoRF is the Python-based code repository for the paper "geoRF: a Geospatial Random Forest". It is a bagged tree ensemble for geospatial data that introduces two split types specifically designed for geospatial features: a diagonal split and a Gaussian split. More information can be found in the paper: full-text access.

Installation

Clone the repository to use the geoRF model.

git clone https://github.com/margotgeerts/geoRF.git

Usage

from geoRF import GeoRFRegressor

# instantiate GeoRFRegressor
regr = GeoRFRegressor(n_estimators=100, n_jobs=-1)

# train the model, give training data and the column indices of the geospatial features
regr.fit(X_train, y_train, geo_features=[0,1])

Also see the example.

Cite us

@article{Geerts2024geoRF,
   author = {Margot Geerts and Seppe {vanden Broucke} and Jochen {De Weerdt}},
   doi = {10.1007/s10618-024-01046-7},
   issn = {1573-756X},
   journal = {Data Mining and Knowledge Discovery},
   title = {GeoRF: a geospatial random forest},
   url = {https://doi.org/10.1007/s10618-024-01046-7},
   year = {2024},
}

Contributions

This is an experimental implementation. If you find any errors, please let me know.

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.