Enlightened Crime Analysis: The Convergence Of Criminology Theories And Machine Learning Techniques

Abstract

The current research aims to the integration of the classical criminological theories and the contemporary advanced machine learning methods for improving crime analysis. This discovers integrating the Rational Choice, Strain and Social Disorganization theories together with the advanced computational models for achieving the general research goals of creating the enhanced, data-driven approach explanation and prediction of criminals’ actions. Techniques from classical theories of crime can offer important understanding of social and psychological roots of crime Yet they are not well suited to address the nonlinear nature of the contemporary, large scale data sets. On the other hand, machine learning can easily detect features in big data but can sometimes miss important nuances given by criminological theory. This research fills that gap showing that the integration of both fields seems to enhance the efficacy and the ethical practice of predictive policing. Outcomes which include model bias, fairness and transparency are discussed with a discussion being made on the ethical use of machine learning in casework. This paper argues that including social science perspectives into the machine learning algorithms helps to improve the correctness of crime predictions and their explanations for the usage of the police.