An Improved LSTM Crime Prediction Model Based on Environmental Factors

Crime prediction serves as a fundamental basis for preventing crime, maintaining social stability, and ensuring public safety. Current crime prediction models commonly suffer from issues such as low accuracy, unstable predictions, or weak generalization capabilities. Therefore, we constructed an improved LSTM crime prediction model based on environmental factors, integrated the data of various environmental factors such as economy, culture and climate, improved the LSTM network structure and optimization method, and enhanced the prediction ability of crime in the region. Empirical evidence shows that the MAPE predicted by the model is 0.0491 and R2 is 0.9229. Compared with the traditional model, the accuracy, stability and generalization ability of the prediction are significantly improved, which can provide a basis for the formulation of crime prevention and control strategies, improve the level of social security management, and promote social stability.