Implementing Geodetector in Python with the py_geodetector Library

This blog post introduces the Geodetector statistical tool, which measures spatial stratified heterogeneity (SSH) and tests the coupling between variables without assuming linear relationships, and explains its three main functions: detecting SSH, testing variable coupling, and exploring interactions.

The official Geodetector software can be downloaded from http://www.geodetector.cn/Download.html , and the Python implementation relies on the open‑source py_geodetector library available at https://github.com/djw-easy/GeoDetector . Users must discretize driving factors before analysis.

After installing the library via pip install py_geodetector , the following Python script demonstrates a full workflow: loading a CSV file, initializing GeoDetector with a response variable and multiple explanatory variables, performing factor detection, interaction detection (with and without relationship output), ecological detection, risk detection, printing results, saving outputs to CSV, and visualizing the results.

import pandas as pd
from py_geodetector import GeoDetector

if __name__ == '__main__':
    file_path = r'2001-1km.csv'
    df = pd.read_csv(file_path, encoding="utf-8")
    gd = GeoDetector(df, "rsei_data", ["tem_data","pre_data","lc_data","dem_data","slope_data","aspect_data"])
    factor_df = gd.factor_dector()
    interaction_df = gd.interaction_detector()
    interaction_df, interaction_relationship_df = gd.interaction_detector(relationship=True)
    ecological_df = gd.ecological_detector()
    risk_result = gd.risk_detector()
    print(interaction_df)
    print(interaction_relationship_df)
    print(ecological_df)
    print(risk_result)
    interaction_df.to_csv(r"2001-1-interaction.csv", index=False)
    interaction_relationship_df.to_csv(r"2001-1-interaction_relationship.csv", index=False)
    ecological_df.to_csv(r"2001-1-ecological.csv", index=False)
    gd.plot(value_fontsize=14, tick_fontsize=16, colorbar_fontsize=14);

The article also includes several illustrative images, a QR code offering free Python learning resources, and links to related Python tutorials and articles.