The Design And Implementation Of Criminal Events Predicting System Based On ARMA

With the development of database technology and the accumulation of crime data records,the public security system database has stored a large amount of crime data,and new crime data are being added annually.But these large number of crime data has not been sufficiently exploited and the theoretical development-related analysis is also not perfect,it cannot provide the scientific decision,how these crime data is sufficiently mined and provide reasonable based research has become a hot spot.As crime data relates with time,accords with the characteristic of time series,adopting autoregressive moving average ARMA(Auto-Regressive Moving-Average)model predicts the future number of crimes.The main parts are as follows.First,introduces the process of ARMA modeling,a sample is provided by Xiamen crime database,using Matlab software finish model,obtains the model order and parameters.A series of steps include data pretreatment,judgment of smooth,autocorrelation analysis,information given order criterion AIC(Akaike information criterion and residual analysis.The data of the next 10 days is acquired depending on the model.The comparison of the predicted value and the actual number of crimes shows that the ARMA model is feasible to predict the number of future crime data.Then,analyzes the system architecture,the system is divided into data layer,business logic and interface layers.Correspondingly,the system database has been designed the most important logical module has been also designed,including Matlab component,stability prediction module,model identification module,model order,parameter estimation,model predictive modules.The Matlab component is laid emphasis on by time series model estimation,such functions as model identification,order determination and parameter estimation are achieved by Matlab algorithm.On this basis,the core functions of crime event prediction based on time series has been implemented.Finally,the crime prediction function has been implemented depending on Java program,facing Xiamen City Public Security Events;Xiamen public security incidents early warning analysis system has been implemented and tested.Results show that the forecasting software based on ARMA model can provide a powerful social crime prevention tools,so that the original large number of historical crime data has been more fully mined,the actual prediction by analysis of historical data shows that the algorithm model is more accurate,the error range is within 20%.