Crime Prediction by Comparing Machine Learning and Deep Learning Algorithms

Preventing crime is important for justice and safety in cities. Using computers to predict crime trends can help make cities safer. Reliable real-time crime prediction is necessary for public safety, but there are still difficulties facing science. Crime rates are influenced by numerous intricate factors. Crime is low in comparison to many predictable events. For computer systems, determining the criminal activity rate, kinds, and hotspots based on historical patterns is both a challenge and an opportunity. The accuracy of machine learning using SVM stacking, KNN, Naïve Bayes, Random Forest, and deep learning using LSTM are compared in this study. This paper presents a comprehensive review of the literature on deep learning and machine learning approaches to crime prediction. It's an important tool for scholars studying the subject because it provides information that law enforcement organizations can use to improve how they prevent and deal with criminal activity. Moreover, the system outlined in the paper offers predictions about potential future crimes, allowing for proactive measures to be taken to prevent them.