Abstract
This paper examined the spatial patterns of crime in a part of Tokyo. 68400 cases reported to the police were classified by 15 crime types, and then Moran's I, one of the indicators of the spatial autocorrelation was calculated using "CrimeStat" software. Moran's I for almost crime types including burglary, theft, snatching indicates significantly positive autocorrelation. To examine the effects of aggregation unit's size on Moran's I, data aggregated by both residential blocks and "chome" districts are used for calculation.The results from two types of spatial weight matrix are also discussed.
