Data mining of police crime records

Morel, Natalia Andreea (2015) Data mining of police crime records. BSc dissertation, University of Portsmouth.

[img] PDF
Restricted to Registered users only

Download (1826kB)

    Abstract

    Crime analysis and predictive policing are extremely current affairs that rely more and more on data mining for effective and efficient ways of obtaining meaningful information from the vast amount of data recorded by the police every day. This project focuses on the design, development and comparative evaluation of predictive models using a variety of algorithms to predict crime shifts and statistics that have the potential of helping law enforcement agencies to apprehend criminals and suppress criminal activity. Three crime recurrence indicators (incidence, prevalence and concentration) are measured and predicted in conjunction with cluster-, grid- and area-based approaches to grouping robberies and anti-social behaviour incidents based on their proximity. From the experiments conducted, it is observed that, out of the three different grouping techniques, the best results for predicting revictimisation are obtained by the grid-based approach, with an accuracy of up to 91% in predicting the crime recurrence indicators.

    Item Type: Dissertation
    Departments/Research Groups: Faculty of Technology > School of Computing
    Depositing User: Jane Polwin
    Date Deposited: 03 Dec 2015 16:33
    Last Modified: 03 Dec 2015 16:33
    URI: http://eprints.port.ac.uk/id/eprint/19073

    Actions (login required)

    View Item

    Document Downloads

    More statistics for this item...