Comprehensive Risk Assessment For Civil Aviation Passengers Based On Machine Learning

With the development of domestic economy,the passenger throughput of civil aviation airport is continuously on a rise,which puts increasing pressure on the security check.Although China's civil aviation security continues to develop in a positive way,the efficiency of the security check still needs to be improved because the focus of the security check are carry-on items instead of people.According to the trend of grading security check development and the problem of lack of quantitative research on characteristics of personal background for passenger risk assessment,a comprehensive risk assessment method for civil aviation passengers based on machine learning is proposed.The works of this paper are listed as follows:Firstly,the data set preprocessing is completed.As for issues of rarely domestic sample cases available for research and algorithm secrecy,this thesis adopts the Transnational terrorism database of open source John Jay & ARTIS to analyze the impact of personal background characteristics on individual risk levels.Besides,the feature attributes of data set are analyzed in detail,an accomplishment of preprocessing process of data set including data cleansing,feature filtering,label construction and oversampling construction of balanced training set,etc.Secondly,the machine learning method is used to select appropriate risk assessment characteristics.Different grades of risk samples are classified by using the method of random forest.It also calculates the importance of the characteristics in the course of classification and the contribution rate of characteristics on the classification by using Gini coefficient.Then part of features whose cumulative contribution rate are over 70% are chose as evaluation indicators of personal characteristics in risk assessment and analyze the impact of those attributes and characteristics on its risks level.Finally,a comprehensive qualitative and quantitative risk assessment method for civil aviation passengers is proposed.Combining the personal characteristic indexes and security information indexes,using expert scoring method as a quantitative assessment method can evaluate the level of passenger risk;based on the qualitative assessment method of persona,analyzing personal characteristics of high-risk passengers to construct persona can provide an intuitive reference for the evaluation of personal risk level during the process of grading the passenger.And this method can serve as an auxiliary evaluation method in the process of security check.