Discovering the Online Game Customer Behavior and Customer Values Using Markov Chain Modeling and Association Rules

• Main Authors: Ying-Chieh Wu, 吳盈杰
• Other Authors: Chun-Te Chen
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/41352739666263236011

碩士 === 華梵大學 === 資訊管理學系碩士班 === 94 === The internet business is changed too fast to understand the interesting change of the customer. The customer relationship management is adopted in the enterprise to provide not only the intimate service to seize the hearts of customers, but also the core competition to satisfy customers’ needs. To raise customer satisfaction and customer royalty, many enterprises are investing many resources to increase in understanding the interesting change of customers and to establish a good relationship with customers. They also attempt to find the higher value customers by the Customer Value Analysis. The player of the on-line game industry is concerning or addressing on the continuous consuming of the user to determine the value of online game firms. Therefore, it must be to understand the consumer behavior. That is the fundamental of issue for finding the value of the game and the value of the customer, making a proper response in time, or satisfying the customer demand. Hence, this paper is proposed to discovery the customer behavior and the customer value from the online game’s log file. It is constructed based on the Markov chain modeling and the association rule mining by using the survival duration time of the customer, the execution time in each game level and the number of login times. The effectiveness of customer segmentation depends on the behavior and the execute time when he/she pass through the some outpost inside the game, such as the major outpost, the minor outpost, or the normal outpost. With these association rules, it can divide the customer behavior into the eight behavior segments. In each segment, the customer is partition into three actors, such as tyro, senior and master dependent on the number of the login times and the execution time. With the operation record of the customer, the behavior transition model is generated by utilizing the Markov chain model. Combined with the RFM model of the customer purchasing history data, the customer value of the customer transition probability matrix is obtained. With these transition matrixes, the probability of behavior changed in each segment and customer migration pathways can be used to estimate the customer value. With the proposed customer behavior model and the customer value, the enterprise of the on-line game industries can monitor the customer state change clearly to gain more benefit and to enhance the customer loyalty for continuous consuming.