Classification for predicting offender affiliation with murder victims

摘要: Abstract The National Incident-Based Reporting System (NIBRS) is used by law enforcement to record a detailed picture of crime incidents, including data on offenses, victims and suspected arrestees. Such incident lends itself the use mining uncover hidden patterns that can provide meaningful insights policy makers. In this paper we analyze all homicide recorded over one year in NIBRS database, classification predict relationships between murder offenders. We evaluate different ways for formulating problems prediction compare four methods: decision tree, random forest, support vector machine neural network. Our results show setting up binary discriminate each type victim–offender relationship versus others good accuracy be obtained, especially method forest approach. Furthermore, our interesting structural insight obtain performing attribute selection using transparent tree models.