This study aims to apply the spatial autocorrelation method to analyze the cluster pattern of road crashes based on geo-information and open data aggregated at the second dissemination area level from 2018 to 2021. The analysis compared the cluster patterns of crashes in different severity levels and different time periods in 359 zones within the study area in Taichung City, Taiwan. Considering the spatial correlation, this study further adopts the spatial regression model to identify the factors that affect the number of crashes.

 Preliminary research results show that under the time scale of the whole day, the number of crashes produced significant clusters of crash patterns. It is shown that the number of crashes, injuries, and the severity indicators have a significantly clustering pattern during the peak hours of the day while all severity levels emerged a significant clustering pattern during the evening peak hours. 

 It is also shown that zones of outliers can be identified by comparing spatial autocorrelation patterns over time periods and with varying degrees of severity. The ongoing work of this research will apply spatial regression model to identify the key factors that affect the number of crashes. The outcomes of this study can be beneficial for local government in prioritizing improvement areas and developing strategies for road safety enhancement.