Geohunter is a python package for parsing and analyzing data from the Overpass API in a pandas-like programming framework. There are similar packages to ours, but our purpose is to create a bridge between data from the OpenStreetMap platform to the geopandas data structures, which offer lots of useful tools for geospatial data analysis.
This package was originally created with the purpose of providing volunteered geographic information to machine learning pipelines. Data from points-of-interest of the city may represent an important source for feature extraction. There are cases where the density of geographic information may help predict a particular variable. For example, consider a set of various points-of-interest.
In our package, we provide some geographic feature extraction procedures, for example with the function util.kde_interpolation
(see result in figure below), based on the Scipy implementation of KDE.
Clone the package, go to the repository directory (where setup.py is) and pip install .
should start the installation.
Please take a look at the /example folder while the documentation is under construction.
Pandas, geopandas, shapely, requests. For visualization (optional), matplotlib and descartes.
We are welcome to your contribution, submit your pull request!
- Adelson Araújo
- João Marcos do Valle
If you use this package in a scientific publication, we would appreciate if you cite the following paper:
Geographic Feature Engineering with Points-of-Interest from OpenStreetMap. Araujo et al., KDIR 2020, p. 116-123.
@conference{araujo2020geographic,
author={Araujo, A. and Valle, J. M. and Cacho, N.},
title={Geographic Feature Engineering with Points-of-Interest from OpenStreetMap},
booktitle={Proceedings of the 12th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management - Volume 1: KDIR},
year={2020},
pages={116-123},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010155101160123}
}