Applying soft cluster analysis techniques to customer interaction information

• Main Authors: DURAN, Randall E., ZHANG, Li, HAYHURST, Tom
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2010
• Subjects: Credit Card; Fuzzy; Cluster; Rand Index; Competitive Learning; Customer Type; Databases and Information Systems; Management Information Systems
• Online Access: https://ink.library.smu.edu.sg/sis_research/6451; https://ink.library.smu.edu.sg/context/sis_research/article/7454/viewcontent/Applying_Soft_Cluster_Analysis_Techniques_to_Customer_Interaction_Information.pdf
• Institution: Singapore Management University

The number of channels available for companies and customers to communicate with one another has increased dramatically over the past several decades. Although some market segmentation efforts utilize high-level customer interaction statistics, in-depth information regarding customers’ use of different communication channels is often ignored. Detailed customer interaction information can help companies improve the way that they market to customers by taking into consideration customers’ behaviour patterns and preferences. However, a key challenge of interpreting customer contact information is that many channels have only been in existence for a relatively short period of time, and thus, there is limited understanding and historical data to support analysis and classification. Cluster analysis techniques are well suited to this problem because they group data objects without requiring advance knowledge of the data’s structure. This chapter explores the use of various cluster analysis techniques to identify common characteristics and segment customers based on interaction information obtained from multiple channels. A complex synthetic data set is used to assess the effectiveness of k-means, fuzzy c-means, genetic k-means, and neural gas algorithms, and identify practical concerns with their application.