Using Machine Learning to Assist Crime Prevention

Drug-related criminal activity is gradually rising in Taiwan and has a significant and negative social impact. This paper proposes a data-driven method based on “broken windows” theory and spatial analysis to analyze crime data using machine mining algorithms and thus predict emerging crime hotspots for additional police attention. The Deep Learning algorithm has been widely applied in several fields, include image recognition and natural language processing. With fine tuning, we find the Deep Learning algorithm provides better prediction results than other methods including Random Forest, and Naïve Bayes for potential crime hotspots. Furthermore, we improve model performance by accumulating data with different time scales. To validate experimental results, we visualize potential crime hotspots on a map, and observe whether the models can identify true hotspots. Finally, we discuss the applicability of this method, and present future research directions.