Analyzing Spatio-Temporal Effects of Social-Economic Factors on Crime

Sebastian Baumbach, Nikita Sharma, Sheraz Ahmed, Andreas Dengel

In: Claus-Peter Rückemann , Technion Yerach Doytsher (Hrsg.). The Tenth International Conference on Advanced Geographic Information Systems, Applications, and Services. International Conference on Advanced Geographic Information Systems, Applications, and Services (Geoprocessing-2018) befindet sich Digital World 2018 March 26-29 Rome Italy Seiten 11-17 Geoprocessing ISBN 978-1-61208-617-0 IARIA, 2018 2018.


Rampant increase in crime incidents has led to the need of crime analysis in greater detail. Existing crime analysis approaches focused on higher spatial granularity (i.e., country or state levels) and consider each data observation independent of each other. However, data can exhibit spatial and temporal relationships among them. Such interrelationships must be taken into consideration if precise crime analysis is intended. Therefore, a two-stage approach is proposed for predicting crime by analyzing its relationship with socio-economic factors: the first stage applies a spatio-temporal analysis on the data and these results are utilized for the spatio-temporal prediction, which forms the second stage. For evaluation, more than 450 different socio-economic factors and crime data for county level in Germany were analyzed. The evaluation results exhibit a mean absolute percentage error of 6.79% for spatio-temporal crime predictions, outperforming traditional regression techniques with an error rate of 37.1% - 37.8%.

geoprocessing_2018_1_30_30109.pdf (pdf, 434 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence