Exploring the Benefits of Data Mining on Juvenile Justice Data

Knowledge discovery and data mining (KDD) is an area of analysis that has emerged from computer science. In this area the emphasis is on the semi-automated extraction of relationships from large datasets rather than the traditional statistical framework of testing relationships to provide evidence for or against a given hypothesis. KDD has proven to be a useful analytical methodology and has been applied in a range of commercial applications such as credit risk assessment, fraud detection and medical survival analysis. Additionally, KDD has had several scientific applications and been applied in biomedical, geospatial and climatic research. Although data mining has proven to be a useful methodology in a number of fields, its use in criminal justice applications and research is in its infancy.

Aims

The purpose of this project is to apply a number of data mining techniques to data of Queensland juvenile court appearances to extract relevant, potentially unforseen, relationships. An analysis procedure allowing significance tests of extracted relationships will be incorporated into the validation process. Extracted relationships may have implications for policies related to juvenile crime prevention such as the target allocation of resources to appropriate intervention programs. The work will be used to test the effectiveness of the method and to develop data mining expertise within JMAG (Justice Modelling @ Griffith). This expertise will be of value to criminal justice organisations responsible for devising justice policies that may have the effect of reducing crime. JMAG currently has strong relationships with Queensland criminal justice agencies. A significant outcome of this work will be a report on discovered relationships for the Department of Communities. In addition opportunities will be used to inform and educate criminal justice agencies on the benefits of KDD. It is anticipated that further consulting/research opportunities in the area may be established.

Outcomes

This project will serve as a case study for the application of data mining to crime data. A technical report will be produced to educate criminological researchers and people in the justice system about data mining, some of the more prevalent techniques that are adopted, the methodology for a data mining analysis, and the results obtained in applying these techniques to crime data. Additionally, documentation will emphasise the ability for Griffith University to provide expertise in the application of the method for Government projects related to the use of data analysis to support evidence-based policy decisions.

Participants

Dr Brett Gray, Assoc. Prof. Anna Stewart, Dr Troy Allard, Dr Andrew Lewis
Griffith University

Industry Participants

Department of Communities
Brisbane