Research On Crime Prediction And Analysis Based On Spation-Temporal Sequence Model

Crime prediction refers to the use of scientific methods to analyze and research the existing historical crime data and various related data that may have an impact on crime,and to predict or determine the state and development trend of criminal phenomena that may occur in a specific time and space in the future.High-precision crime prediction can provide important scientific guarantees for the formulation of crime prevention strategies and tactical measures.This paper proposes a spatial weight matrix construction method based on taxi data.On this basis,the spation-temporal autoregressive moving average model is used for crime prediction.In addition,the spation-temporal autoregressive moving average model cannot fit the non-linear features in the spation-temporal sequence of crime.This paper proposes a neural spation-temporal prediction model based on neural networks.The main work of this paper is as follows.(1)With the development of transportation and frequent human travel activities,the proximity relationship between regions is not only the proximity in spatial location.This paper combines a taxi trip data and map structure information to design a spatial weight matrix construction method.This method redefines the distance relationship between regions to better reflect the nature of the crime's proximity to space.(2)Based on the designed spatial weight matrix,use the STARMA model for crime prediction.Experiments verify the effectiveness of the algorithm.(3)Since the spation-temporal autoregressive moving average model is a linear model,it cannot capture the non-linear features in the spation-temporal sequence of crime.Based on the designed spatial weight matrix,this paper proposes a neural space-time prediction model based on neural network.Experiments show that the model can improve the accuracy of crime prediction to a certain extent.Finally,the paper summarizes the full text and puts forward some prospects for further research.