Predicting crime in place and time : comparing models, spatial context, data and interpretations

Korte inhoud:This doctoral research aims to contribute to a more context-sensitive and methodologically informed application of supervised machine learning (ML) in predicting where and when crime is most likely to occur. Rather than focusing solely on maximizing predictive accuracy, this research examines how spatial context and methodological decisions influence both the performance and interpretability of spatiotemporal crime predictions derived through ML modeling. Through comparative research, this research evaluates how different ML algorithms perform across different European cities, and how the selection of microgeographic units and the use of open versus proprietary (big) data sources influence crime prediction. It also compares how explainable AI (XAI) techniques can make complex ML models more transparent and interpretable. This doctoral research advocates for a closer integration of data-driven and theory-driven research to advance criminological (theoretical) knowledge and evidence-based crime prevention.

Jaar van publicatie:2025

Toegankelijkheid:Embargoed