Deveolping methodologies for customer classification in CRM : decision trees and collaborative recommendations

Abstract

Customer classification has been become one of important parts of CRM. Moreover, classification has provided many benefits to CRM implementation such as reducing marketing cost and time, improving the accuracy of marketing campaign, through grouping similar customers or assigning a class to customers, based on customers’ preference. Main research framework in CRM is composed of three key concepts: Customer Selection, Customer Targeting, and Relationship Marketing. Customer Selection executes searching customers who have high likelihood to purchase or high possibility to churn. Customer targeting performs an analysis that a marketer can know what proper marketing methods is for individual customer. Customer classification method is necessarily used in Customer Selection, and Customer Targeting at same manner, which extract a customer grouping with similar preference or rules to explain a characteristics of a customer grouping. Finally, Relationship Marketing concentrates on a customer life-time value that represents total purchasing monetary value all during life time. For leveling up of life time value of customers, recommender systems for up-selling and cross-selling are generally used to. A basic principle of recommender systems, which is personalization techniques that apply data analysis to the problem of helping customers find products, is similar to that of customer classification. In this thesis, for classifying customers effectively, we develop methodologies for two issues but other researcher have not been solved yet in CRM field. First issue considers the way in which a customer’s purchase sequence evolves over time, for the purpose of enhancing the quality of recommendations. Second issue is a methodology for preventing a bushy classification tree, in the case of segmenting customers in Internet shipping mall that have a lot of noise information. Firstly, In general, the purchasing sequence of a customer in the database who has made repeat purchases...