2026-10-07T02:42:09Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33944155 2026-09-20T01:04:20Z category_29887 item_type_3 month_year_09_2026

Detection of special-types of spatiotemporal heterogeneity in spatiotemporal autoregressive varying coefficient models jing Chen (23747097) Spatial statistics Local autoregressive geographically and temporally weighted regression Spatiotemporal heterogeneity Bootstrap method <p dir="ltr">Local autogressive geographically and temporally weighted regression (LARGTWR) models have been demonstrated to be an effective tool for simultaneously capturing the spatiotemporal heterogeneity of autocorrelation in the response variable and regression relationship. In this model, both the autoregressive and regression coefficients are assumed to be varying over both space and time. In many practical situations, however, constant, only spatial varying, only temporally varying coefficients are more likely to exist. Therefore, it is necessary to develop newmethods for identifying the special types of the autoregressive and regression coefficients in order to give a deep understanding of spatiotemporal characteristics of the autocorrelation in the response and regression relationship. In this paper, we first extend the average-based estimation in the conventional geographically and temporally weighted regression (GTWR) model to the case of LARGTWR models and obtain the estimates of the autoregressive and regression coefficients. Then, two statistics based on the residual sum of squares are constructed on one hand for identifying whether the autoregressive coefficient is constant, only spatially varying or only temporally varying, and on the other hand for ascertaining if there exist any regression coefficient which remain constant, only spatially varying, or only temporally varying. Further, a bootstrap method is employed to approximate the 𝑝-values of the tests. Simulation experiments are conducted to evaluate the performance of the test methods and the results demonstrate that the proposed test statistics have high accuracy and satisfactory power in inferring the special types of both the autoregressive and regression coefficients in the LARGTWR model. A real world data set is analyzed to show the application of the proposed methods.</p> 2026-09-20T01:04:20Z Dataset Dataset 10.6084/m9.figshare.33944155.v1 https://figshare.com/articles/dataset/Detection_of_special-types_of_spatiotemporal_heterogeneity_in_spatiotemporal_autoregressive_varying_coefficient_models/33944155 CC BY 4.0