This paper builds a framework about the crime data mining of on intelligence analysis and decision command based on the principle of prevention and control in public security. First, we can get the standardized crime information according to the preprocessing of crime database and space encoding. Using the related methods used in cluster analysis, classification and association analysis, we can get such information as the spatial-temporal risk distribution of crime, targeted people features and modus-operandi. Finally, by mining the data of actual theft cases in Beijing, it is proved that data mining methods could play significant role in crime intelligence analysis.