Crime Rate Prediction Using Combined Arima and LSTM

Urbanization created a plethora of social problems, and one ubiquitous issue in cities worldwide was crime.Police databases accumulated vast amounts of data that could be analyzed to mitigate crime rates.The analysis of criminal activity and prediction of the number of crimes remained one of the most interesting problems for researchers.For a developing country like India, it was not uncommon for people to hear about crimes occurring frequently.With the rapid urbanization of cities, there was a constant need to be aware of our surroundings to avoid unfortunate incidents.In an effort to observe crime rates and prevent them, this work attempted to analyze crime rates using a hybrid prediction method.A hybrid model based on deep learning methods, integrating an autoregressive integrated moving average (ARIMA) model and a long short-term memory (LSTM) model, was proposed to enhance the accuracy of crime rate prediction.The analysis provided a comprehensive guide to the crime rate analysis of model parameters concerning performance in predicting crime rates, with accuracy calculations derived from comparing supervised classification machine learning algorithms.