GVNNWR: Geographically and Vertically Neural Network Weighted Regression for Three-dimensional Spatial Non-stationarity Modeling

This repository provides complete experimental implementation methods and evaluation result generation for the Geographically and Vertically Neural Network Weighted Regression, which develops a 3D Spatial Proximity Neural Network (SPNN3D) that integrates multi-dimensional distance metrics within a neural framework to model nonlinear spatial proximity and finally be combined with GNNWR to better characterize complex spatial non-stationarity in three-dimensional geographical space. This contains the code of model implementation and model evaluation, as well as simulation datasets and real case datasets.本代码仓库提供了地理垂直神经网络加权回归（Geographically and Vertically Neural Network Weighted Regression，简称GNNWR）的完整实验实现方案与评估结果生成流程。该方法构建了三维空间邻近性神经网络（3D Spatial Proximity Neural Network, SPNN3D），其在神经网络框架内融合多维度距离度量以建模非线性空间邻近关系，并最终与GNNWR结合，从而更精准地刻画三维地理空间中的复杂空间非平稳性。本仓库包含模型实现与模型评估的代码，以及模拟数据集与真实案例数据集。