Identification and Classification of Alcohol-Related Violence in Nova Scotia Using Machine Learning Paradigms

Abstract

A significant improvement in big data analytics has motivated the radical change in the scientific study of crime and criminals. In terms of criminal activities, it has been observed that alcohol has a great influence in most of the cases. The main goals of our research are to analyze different types of violence happening in Nova Scotia and to apply machine learning techniques to model the relationships between alcohol consumption and violence. In many machine learning algorithms, it is assumed that, the training and testing data must be in the same distribution and feature space. Because of limited amount of Nova Scotia criminal activity data, the need of transfer learning arises which helps to gain knowledge from different domains. The results of our studies show a very satisfactory classification performance on Nova Scotia data.