Extending Getis-Ord Statistics to Account for Local Space-Time Autocorrelation in Spatial Panel Data

• Published in: The Professional geographer 2020-07, Vol.72 (3), p.411-420
• Main Authors: Wang, Zheye, Lam, Nina S. N.
• Format: Article
• Language: English
• Subjects: and G; análisis exploratorio de datos del espacio-tiempo (ESTDA); autocorrelaciãn del espacio-tiempo local; Autocorrelation; Clustering; Data analysis; Data processing; espacio-tiempo G; exploratory space-time data analysis (ESTDA); local space-time autocorrelation; Longitudinal studies; matriz de peso contemporánea del espacio-tiempo; matriz de peso rezagada del espacio-tiempo; Panel data; space-time contemporaneous weight matrix; space-time G; space-time lagged weight matrix; Spatial analysis; Spatial data; Statistical analysis; Statistics; Visualization; Weight; y G; 局部时空自相关; 时空G; 时空同期权重矩阵; 时空数据探索分析 (ESTDA); 时空滞后权重矩阵
• ISSN: 0033-0124; 1467-9272
• DOI: 10.1080/00330124.2019.1709215

Space and time are both crucial characteristic dimensions of geographic events and phenomena. Although exploratory spatial data analysis (ESDA) can be used to visualize and summarize complex spatial patterns, it has limitations in capturing the temporal dynamics of geographic features. Efforts have been made to incorporate the time dimension into ESDA techniques to detect space-time clustering or trends. Localized space-time statistics that could help in exploratory space-time data analysis (ESTDA), however, are still lacking. Focusing on spatial panel data, our work extended Getis-Ord and statistics using a space-time contemporaneous weight matrix and a space-time lagged weight matrix to account for local space-time autocorrelation. Two applications in this article show that the newly developed method can be used to summarize space-time patterns from spatial panel data, identify changes of landscape more consistently, and lend the results readily to visualization.