Leveraging Artificial Intelligence for Predictive Policing: Ethical, Legal, and Technical Challenges

Predictive policing has an impact on crime prevention through the utilization of data analytics machine learning (ML) and AI. It aims to identify potential crime hotspots make the most of police resources, and reduce criminal activity by examining demographic information historical crime trends, and current data. This study will focus on machine learning techniques to anticipate criminal behavior. We aim to discover the most effective approach to train a system that can foresee unlawful acts in major urban areas. The project intends to provide the Police Department with specific crime forecasts helping them allocate their resources to areas where crimes are likely to occur. In order to answerability the "lies and" column, this study employed ML methods to forecast the probability of an arrest for alleged crime. I created four ML models: SVM DT, XGBoost and ANN. The research team evaluated these models on the Police Department's CLEAR (Citizen Law Enforcement Analysis and Reporting) system, to over 7,0 million records history. The ANN got the best accuracy (95.9%) among the four models. The paper also helps to understand in which ML techniques crime dataset from large cities behave differently.

Перевод пока недоступен