New research, by University of Waikato Honorary Research Associate Dr Sophie Curtis-Ham, could help police resources go further by automating laborious analysis while eliminating subjectivity and human error.

Dr Curtis-Ham has developed a new algorithm that automates part of the geographic profiling process to make the process more efficient for analysts in investigative police work.

Geographic profiling involves analysing geographic aspects of crimes to prioritise places or people for the investigation to focus on.

The algorithm prioritises suspect lists for further investigation by analysing crime locations and areas familiar to suspects. It has already been tested on thousands of solved cases and offers promisingly high accuracy.

"In the solved cases we tested it on, the algorithm consistently placed the actual offender at or near the top of the suspect list," says Dr Curtis-Ham.

Dr Curtis-Ham says it was in her day to day work for police that she recognised the need for an algorithm to make geographic profiling faster and more accurate for analysts like herself:

"Geographic profiling is a valuable tool for investigative work, but I see a real need to reduce the time required for manual analysis of geographic data. And in some cases the sheer volume of data makes it untenable for manual analysis.

"I also wanted to remove the subjectivity that can make the process less effective."

Dr Curtis-Ham's algorithm, the Geographic Profiling: Suspect Mapping and Ranking Technique (GP-SMART), was created as a practical application of her thesis with the University of Waikato.

To build the algorithm, Dr Curtis-Ham worked with local burglary, robbery, and non-family-based sexual assault crime data to ensure it told a similar story to off-shore research.

She then built on the prior research by exploring which places people are more likely to commit crimes near.

The technique uses suspects' home addresses and locations of prior offences. But what makes Dr Curtis-Ham's approach different from the traditional geographic profiling process is her inclusion of wider activity nodes for individuals, including work or school, and prior arrest locations.

"I could see from my experience working in the field of investigative analysis for police that data on suspects' activity nodes could produce faster, more reliable results."

She is now testing the tool with real crimes and cold cases in her work at the New Zealand Police's Evidence-Based Policing Centre.

"The result of Sophie's research is a body of work that is already influencing international researchers and will certainly help police resources go further," says Professor Devon Polaschek from the University of Waikato.

Dr Curtis-Ham was awarded the University of Waikato 2022 Koko Kairangi Award for Best Doctoral Thesis in recognition of this real-life, practical research.