The Research Of The Segmentation Of Airline Passengersbased On Booking Behavior

In recent years, with the rapid development of domestic economy, the number of the civil aviation passengers are rapidly increasing, and the domestic civil aviation has come into fast development mode. In order to deal with the fierce competition of the civil aviation market, it’s necessary for airlines to analyze the travel preference for different passenger groups and formulate the corresponding competition strategy. Therefore, this paper uses the information of customer booking records as data source, adopting the clustering technology, and then analyzes the travel preference of air passengers on the basis of the effective customer segmentation.Differing from the numerical data for the analysis of the traditional clustering algorithm, this paper analyzes the data of air passengers booking behavior, and the specificity lies in the following reasons. Firstly, the data source is the mixed type dataset which contains numerical data and the categorical data; Secondly, the amount of data is enormous and distributed stored in the airlines; Therefore, this paper improves the existing clustering algorithm so as to deal with the mixed-data clustering for a single airline, and analyzes the travel preferrence, then design a distributed clustering algorithm so as to analyze the travel preference for air passengers from the global perspective. Therefore, this paper mainly includes the following two aspects:(1) Based on the booking record of passengers, this paper views the passengers partition as the mixed data segmentation, and then takes the k-prototypes algorithm to achieve the airline passengers segmentation. At the same time, because some of the attributes of the booking data for passengers are discrete, category numerous and semantic fuzzy, with the aid of the civil aviation domain knowledge, this paper uses the data mining method based on the domain knowledge to deal effectively with these data, simplifying the category of the data attributes. At the meantime, by building the quantitative calculation model of the value of the passengers, the method proposed by this paper improves the effectiveness of the passengers segmentation and the visualization of the analysis for the passengers’ travelling behavior and preference.(2) To deal effectively with the mixed attributes of these mass data sets, this paper proposes the Domain based Parallel K-prototypes algorithm(DPKP) by extending the K-prototypes algorithm, combining with the domain knowledge to run this algorithm in parallel manner. This method makes the passengers segmentation and data analysis completed on each site, which not only improves the operation efficiency of the algorithm but protects the business privacy of airlines when achieving the segmentation among airline passengers from different airlines. Experimental results show that the proposed clustering algorithm is available for the airline passengers segmentation, improving the accuracy of clustering results and the efficiency of clustering time.Finally, this paper uses the customer data resource of some domestic airline and the clustering algorithm proposed by this paper to build a customer segmentation model Through analyzing the result of customer segmentation, this paper analyzes the travel preference of different customer groups and supplies marketing strategies to the airlines.