Who is most likely to offend in my store now? Statistical steps towards retail crime prevention with Auror

Authors

DOI:

https://doi.org/10.21914/anziamj.v57i0.10507

Keywords:

Retail theft, Prediction

Abstract

Auror is establishing itself both locally and internationally as a leader in retail crime solutions. In mid-2015 a study group of mathematicians and statisticians teamed up with Auror to analyse data from the first two and a half years of their venture to identify and prevent retail theft. The aim was to explore methods for nominating the top ten individuals most likely to offend in a particular store at a particular time. Various methods were employed to explore the relationships between retail crime incidents, including generalised linear models, regression trees and similarity matrices. The relationships identified were then used to inform predictions on individuals most likely to reoffend. The focus of the current analysis is to model the behaviour of reoffenders. At the time of the study group the project was still in the early phases of data collection. As data collection proceeds, prediction methods will likely give better and better intelligence to aid crime prevention efforts. References

Author Biographies

Barry William McDonald, Massey University

Senior Lecturer, Institute of Natural and Mathematical Sciences

Lisa Hall, University of Canterbury

PhD Student, Department of Mathematics and Statistics

Xingyou Philip Zhang, Callaghan Innovation

Computational and Data Sciences

Published

2017-08-13

Issue

Vol. 57 (2015)

Section

Proceedings of the Mathematics in Industry Study Group