Cri-Astrologer: Predicting Demography of Involved Criminals based on Historical Data

Abstract

Because of the rapid advancement in computer technology, police enforcement agencies are now able to keep enormous databases that contain specific information about crimes. These databases can be utilized to analyze crime patterns, criminal characteristics, and the demographics of both criminals and victims. Through the application of various machine learning algorithms to these datasets, it is possible to generate decision-aid systems that can assist in the conduct of police investigations. When there is a large amount of data accessible, several data-driven deep learning approaches can also be utilized. Within the scope of this investigation, our primary objective is to create a tool that may be utilized during the standard investigative process. To forecast criminal demographic profiles using crime evidence data and victim demographics, we present a deep factorization machine-based DNN architecture. We evaluate the performance of our architecture in comparison to that of traditional machine learning algorithms and deep learning algorithms, and we provide our findings in a comparative study.