Prediction of Crime Patterns using the Spatio-Temporal feature relations

Abstract

There are many openly available Crime datasets provided by the government agencies. This theses work attempts to learn the spatiotemporal relations between the different features from the Baltimore Crime and Arrests dataset respectively. The goal is to make a comparative study by exploring different Machine Learning algorithms and finding the patterns that emerge from these learning models. The features to be predicted from the Crime dataset are Crime Code and Premise and the features to be predicted from the Arrest dataset are the Incident Offence and the Arrest Location. These analyses can help us in understanding the effectiveness of using Machine Learning techniques when it comes to finding the patterns related to Criminology.