The forensic information identification based on machine learning algorithms

Abstract

Investigating crimes in selected areas, studying their occurrence and appropriate graphical visualization may be an important supportive information for police units. Nowadays, the available IT tools can improve the work of law enforcement agencies. For this purpose, in this work, we propose an exploration approach which allows for analyzing crime events that are recorded in the city of Baltimore, Maryland, USA. Based on the collected data, seven types of offenses are recorded. First, they are analyzed in terms of time verifying whether any type of event depends on the day of the week or a period of the day. Then the distribution of the crimes is examined by spatial clusterization and kernel density estimation methods. As a result of the analysis, it is shown when and where the citizens of the considered city are subject to the highest crime rate.