Behavior Analysis Of Terrorist Based On Regional Data

In recent years,the global terrorism cases have occurred frequently,and the counter-terrorism situation at home and abroad has become extremely serious,and the anti-terrorism work has become the first important task of the public security departments.Since 2014,China’s terrorism activities have entered a new round of activity,Xinjiang terrorist activities show an obvious spillover trend.Shandong,as the southem gate of Beijing,connects the north and south strategic hub,and the important part of the integration of Beijing,Tianjin and Hebei has a special geographical position.Objectively,it provides convenient conditions for the infiltration and spread of terrorist forces,and the situation related to terrorism is grim.However,the traditional after-attack has been unable to meet the needs of the reality of anti-terrorism,and the traditional integral model has doubts about the accuracy and scientific nature of weight setting..Therefore,it is particularly important to study the behavior of terrorist entities and improve the weight calculation method of the traditional integral model.In the aspect of algorithm selection,this paper mainly improves the weight calculation of the traditional integral model.The logical regression algorithm(logistic regression),which introduces machine learning,has unique advantages for the problem of two classification at the same time.The weight of the traditional integral model is generally determined by the subjective experience of the scout,and it is controversial whether the score is appropriate or not.Its algorithm is relatively simple.It is not as complex as random forest,SVM,neural network,GBDT and other classification algorithms,but it is also more targeted and effective.It is more mature and accurate.In the aspect of model parameter estimation,the maximum likelihood estimation method is selected to estimate the model parameters,and the optimization problem with logarithmic likelihood function is studied.in this paper,the gradient drop method is used to calculate the maximum likelihood estimation of weight.In the aspect of the data source,some of the currently known partial characteristic data in the abnormal behavior of the fear entity are selected,and the tagging is carried out to facilitate the increase or decrease of the later feature,and the partial feature is selected to construct the training data set.In the aspect of algorithm verification,the mainstream Python programming language is selected to verify the effectiveness of the logical regression algorithm from the aspects of data preparation,preprocessing,training,learning and so on.At the same time,the traditional integral model of artificial assignment weight is calculated with the same data set,and the results are less accurate and scientific than the logical regression algorithm.