The K-function analysis of space-time point pattern on road network

The paper proposes one space-time version of network K-function for detecting space-time distribution pattern of incidents occurring on road network. It extends existing network K-function for purely network space point process data to the network space-time setting. Two Monte Carlo simulations are implemented in the experiment, one is to detect space-time clustering of load-unload points, and the other is to test space-time interaction. The experiment results show that there is a significant evidence of clustering and space-time interaction at several space and time scales. The real application indicates the proposed network K-function's potential of analyzing space-time point process.