Data Mining Techniques for Autonomous Exploration of Large Volumes of Geo-referenced Crime Data

We incorporate two knowledge discovery techniques, clustering and association-rule mining, into a fruitful exploratory tool for the discovery of spatio-temporal patterns. This tool is an autonomous pattern detector to reveal plausible cause-effect associations between layers of point and area data. We present two methods for this exploratory analysis and we detail algorithms to effectively explore geo-referenced data. We illustrate the algorithms with real crime data. We demonstrate our approach to a new type of analysis of the spatio-temporal dimensions of records of criminal events. We hope this will lead to new approaches in the exploration of large volumes of spatio-temporal data.