Abstract: In order to rapidly analyze the role of airline agents in ticket sales market and make corresponding cooperation plan as well as sales plan for different types of agents, aiming at the subjectivity of traditional classification methods and the difficulty of centralized system framework in massive data clustering and analyses, distributed Canopy-K-means algorithm is proposed in airline agents clustering, combining clustering results with actual market situation to obtain the agent’s role in market. Experimental results show that the final clustering result is in line with the actual evaluation results of relevant agents, and has practical significance. Finally, according to the clustering results, feasible suggestions for airlines are provided, helping airlines to manage sales agents.

Key words: distributed cluster, clustering algorithm, airlines, agent