HGGNN data and code

Abstract Urban area data are strategically important for public safety, urban management, and planning. Previous research has attempted to recover values from unsampled areas, while minimal attention has been paid to the values of irregular areas, which are naturally formed by road networks or administrative areas. To address this problem, this study proposes a hierarchical geospatial graph neural network model based on the spatial hierarchical constraints of areas. The model first characterizes spatial relationships between irregular areas from different spatial scales. Then, it aggregates information from neighboring areas with graph neural networks, and finally, it imputes missing values in fine-grained areas under hierarchical relationship constraints. To investigate the performance of the proposed model, we constructed a new dataset consisting of the urban statistical values of irregular areas in New York City. Experiments on the dataset showed that the proposed model outperformed the baselines by 30.34%, 14.43%, and 14.47% in RMSE, MAE, and MRE, respectively. This research enriches datasets on spatial missing value research and provides a methodological reference for the completion and prediction of a wide variety of geographic big data. It can be used in numerous spatial decision applications, including traffic management, public safety, and public resource allocation. Keywords: urban area, spatial prediction, hierarchical constraint, spatial interpolation This is the code and data of "A hierarchical constraint-based graph neural network for imputing"

摘要：城市区域数据对于公共安全、城市管理与规划具有战略重要性。过往研究已尝试补全未采样区域的数值，但针对由道路网络或行政区划自然形成的不规则区域的相关研究却鲜有关注。为解决这一问题，本研究提出一种基于区域空间层级约束的层级式地理空间图神经网络（Graph Neural Network, GNN）模型。该模型首先刻画不同空间尺度下不规则区域间的空间关联关系，随后通过图神经网络聚合邻域区域的信息，最终在层级关系约束下对细粒度区域的缺失数值进行插补。为验证所提模型的性能，我们构建了一个包含纽约市不规则区域城市统计数值的全新数据集。在该数据集上的实验结果表明，相较于基准模型，所提模型在均方根误差（Root Mean Square Error, RMSE）、平均绝对误差（Mean Absolute Error, MAE）以及平均相对误差（Mean Relative Error, MRE）上分别提升了30.34%、14.43%与14.47%。本研究丰富了空间缺失值研究的数据集资源，为各类地理大数据的补全与预测提供了方法学参考，可应用于交通管理、公共安全、公共资源配置等诸多空间决策场景。关键词：城市区域、空间预测、层级约束、空间插值 本代码与数据对应论文《A hierarchical constraint-based graph neural network for imputing》