pygwmodel: Python wrappers for building geographically weighted models

Overview

This package includes techniques from a particular branch of spatial statistics, termed geographically weighted (GW) models. GW models suit situations when data are not described well by some global model, but where there are spatial regions where a suitably localized calibration provides a better description.

The goal of pygwmodel is to provide conscious and easy-to-use user interface to high-performance C++ implementations of GW models (see libgwmodel) based on GeoPandas. We believe with the newly designed interfaces and the underlying C++ core, users will get fluent experiences.

Installation

We highly recommend installing this package in a conda environment, especially on Windows.

conda install armadillo gsl openblas catch2 numpy geopandas nanobind scikit-build-core
git clone https://github.com/GWmodel-Lab/pygwmodel.git
pip install ./pygwmodel

On Windows, it is very important to use the OpenBLAS library instead of that bundled with Python to avoid segmentation fault. To do so, set the environment variable CMAKE_ARGS:

# powershell
$Env:CMAKE_ARGS="-DBLA_VENDOR=OpenBLAS"
pip install .

In command line prompt, it can be done like this:

set CMAKE_ARGS="-DBLA_VENDOR=OpenBLAS"
pip install .

Or using the following line to install this package:

pip install . --config-settings=cmake.args=-DBLA_VENDOR=OpenBLAS

Getting started

from pygwmodel import GWRBasic, BandwidthWeight, CRSDistance
algorithm = GWRBasic(data, y, x,
                     weight=BandwidthWeight(36.0, adaptive=True),
                     distance=CRSDistance()).fit()

Multiscale GWR (MGWR) assigns a separate bandwidth to each predictor:

from pygwmodel import GWRMultiscale, BandwidthWeight

n_var = 4  # intercept + 3 predictors
weights = [BandwidthWeight(36.0, adaptive=True) for _ in range(n_var)]

algorithm = GWRMultiscale(data, y, x, weights=weights).fit()
print(algorithm.diagnostic)

For full usage, please see the unit tests in test directory and the documentation <https://gwmodel-lab.github.io/pygwmodel/>_.

Related work

This package is based on a pure C++ library --- libgwmodel. This library implements all models, and pygwmodel just calls this package by translating inputs and outputs. Besides, it also provides some handy functions for the convenience of python users.

Getting help

If you encounter a bug, please create an issue here. It would be better for us if a minimal reproducible example is also provided.