Enhancing Urban Data Analysis: Adaptive Partitioning Framework for Multidensity Regions

As the fundamental units in spatiotemporal data research, appropriate analytical region partitioning directly affects the accuracy of the following analysis. This article proposes a framework for basic region partitioning, named adaptive hotspot area (AHA), aiming to overcome the issues posed by urban data with diverse density distributions and research backgrounds. The framework first introduces the adaptive hotspot detection of multidensity distribution data. Subsequently, AHA establishes basic analysis elements with hotspots as the core control points to ensure that event points are clustered and homogenous within the region and that there is heterogeneity between regions. Using crime data from New York City, case studies apply the AHA framework for region partitioning. Comparative analyses with other methods include geographic unit, spatial autocorrelation, and regression analyses. Results show AHA accurately reveals local spatial clustering and object heterogeneity, offering advantages in predictive modeling and event pattern studies.