Forecasting Crime Event Rate with a CNN-LSTM Model

In India, the crime percentage is growing day by day. It is essential to develop different modern advanced tools and techniques to predict the rate and time of the crime events in advance. This prediction will enable the police to improve the monitoring/vigilance and strengthen intelligence in the particular district to avoid such crime events. There are several Spatiotemporal statistical methods are used to predict such events in the past. Forecasting crime event rate prediction is a central part of setting a prediction approach or taking suitable timely action to reduce the crime rate. Additionally, using this Long short-term memory (LSTM) that one can analyze the relationships among long-term data utilizing its functions. Therefore in this work, we attempted to forecast the crime rate using the CNN-LSTM model. For this research, we utilized the crime dataset taken from the NCRB for three years. We chose four features: murder, rape, theft, and offenses against property. Initially, we use CNN to excerpt the attributes from the dataset and we used LSTM to forecast the crime rate. During the experiments, we found that the CNN along with the LSTM model could provide a trustworthy crime predicting method with high forecast accurateness. This method is a new exploration idea for crime rate forecasting as well as a good prediction technique.