Influence factor identification analysis method and system based on multi-scale geographic space relationship

By using a multi-scale geospatial relationship influencing factor identification and analysis method, and utilizing fully connected neural networks and SHAP values, the shortcomings of traditional models in multi-scale spatial heterogeneity and high-dimensional data processing are solved, and the accurate capture and efficient interpretation of nonlinear relationships are achieved.

CN120951166APending Publication Date: 2025-11-14NANJING UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511066030.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-31
Publication Date
2025-11-14

AI Technical Summary

Technical Problem

Traditional geographic weighted regression models struggle to accurately capture multi-scale spatial heterogeneity, cannot effectively handle nonlinear relationships and high-dimensional multi-source spatial data, and have limited interpretability.

Method used

A multi-scale geospatial relationship influencing factor identification and analysis method is adopted. By rasterizing the dependent and independent variables, a fully connected neural network model is constructed, and SHAP values ​​are used for multi-scale interpretability analysis to identify the contribution of influencing factors.

Benefits of technology

It significantly improves the ability to capture nonlinear relationships, robustly handles high-dimensional multi-source data, provides refined quantitative interpretations, and enhances the model's real-world fit and computational efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120951166A_ABST
    Figure CN120951166A_ABST
Patent Text Reader

Abstract

The invention discloses an influence factor identification analysis method and system based on a multi-scale geographic space relationship, and belongs to the technical field of geographic information processing. The method comprises the following steps: digitalizing a dependent variable space range, and recording a pixel set of each surface element by rasterizing surface elements serving as dependent variables; independent variable multi-scale feature value extraction: converting an independent variable layer into a feature grid, and generating a group of multi-scale influence factor feature values for each surface element; constructing a structured data set; constructing and training a neural network geographic regression model; and calculating an SHAP value for the prediction result of each surface element in the data set so as to quantitatively explain the contribution degree of each influence factor under different scales. According to the method, multi-source heterogeneous spatial data can be effectively processed, multi-scale environment characteristics are constructed to capture the spatial scale effect, and depth and quantitative attribution interpretation for each analysis object is provided for a black box of the model by using an SHAP method.
Need to check novelty before this filing date? Find Prior Art