The development of policing strategies and the implementation of measures for crime control and reduction depend heavily on crime prediction. Today, machine learning is the most popular prediction technique. However, there haven't been many studies that thoroughly contrasted various machine learning approaches for criminal behaviour prediction. substantial seaside city in the southeast In this study, The ability of several machine learning algorithms to forecast crime is assessed using public property crime data from China from 2015 to 2018. The LSTM model appears to outperform KNN, Random Forest, Support Vector Machine, Naive Bayes, and Convolution Neural Networks based only on results from historical crime data. As a result, it is advised that characteristics linked to criminological theories and knowledge of previous crimes be used to predict future crime. Not all machine learning methods for predicting crimes are equally effective.