Offenders Clustering Using Fcm And K-means

Abstract

One of the most applicable and successful methods to provide security in society is to use data mining techniques to recognize patterns of crimes. Data mining is a field that discovers hidden patterns of large amount of data in large data bases, and also extracts useful knowledge in every field which uses it. Clustering is a technique of data mining that divides data points into many groups so that the members of each group have the most similarity and the members from different groups have the least similarity. In this paper we cluster 100 offenders according to crime they have committed, using Fuzzy C-Means and K-Means algorithms in Matlab and Weka environments. Then we studied the intersections in efficient elements in crime occurrence in each cluster. We obtained interesting results coincided our real data. Hence we have created a pattern which is able to detect crime with considering other attributes, and reversely. It is clear that these detections can help to decrease the effects of crime. Note that Fuzzy CMeans algorithm has provided more accurate results in comparison with K-Means algorithm, because of considering fuzzy point of view and natural uncertainty in the real world.