Observing Customer Segment Stability Using Soft Computing Techniques and Markov Chains within Data Mining Framework

Observing Customer Segment Stability Using Soft Computing Techniques and Markov Chains within Data Mining Framework

Abdulkadir Hiziroglu (Department of Management Information Systems, Yıldırım Beyazıt University, Ankara, Turkey)
Copyright: © 2015 |Pages: 17
DOI: 10.4018/ijissc.2015010104


This study proposes a model that utilizes soft computing and Markov Chains within a data mining framework to observe the stability of customer segments. The segmentation process in this study includes clustering of existing consumers and classification-prediction of segments for existing and new customers. Both a combination and an integration of soft computing techniques were used in the proposed model. Segmenting customers was done according to the purchasing behaviours of customers based on RFM (Recency, Frequency, Monetary) values. The model was applied to real-world data that were procured from a UK retail chain covering four periods of shopping transactions of around 300,000 customers. Internal validity was measured by two different clustering validity indices and a classification accuracy test. Some meaningful information associated with segment stability was extracted to provide practitioners a better understanding of segment stability over time and useful managerial implications.
Article Preview

Motivation And Background

The main idea of segmentation or is to group similar customers. A segment can be described as a set of customers who have similar characteristics of demography, behaviours, values, and so on (Nairn and Berthon, 2003, Bailey et al., 2009). In fact, one of the most valuable pieces of information based on which a segment can be customers’ behavioural characteristics, especially past customer purchases and value-oriented attributes (Bayer, 2010, Kim et al., 2006, Wind and Lerner, 1979). In fact, customer analytics related technological advances have facilitated performing segmentation studies based on those characteristics (Bailey et al., 2009).

This study focuses on two main issues regarding customer segmentation. The first issue is related to the difficulty of obtaining homogenous segments. For customer segmentation, a wide variety of data analysis techniques, such as cluster analysis (Ghazanfari et al., 2008; Hruschka et al., 2004, Li et al., 2009, Liu and Shih, 2005, Wang, 2009, Wang et al., 2008; Xia et al., 2010), clusterwise regression (Desarbo et al., 2008), AID/CHAID (Aravamudhan, 2011; Gil-Saura and Ruiz-Molina, 2008, Jonker et al., 2004), multiple regression (Suh et al., 1999), discriminant analysis (Tsiotsou, 2006), latent class structure (Wu and Chou, 2011) and sophisticated soft computing techniques including such as fuzzy-oriented approaches (Crespo and Weber, 2005, Hu and Sheu, 2003, Kaymak, 2001, Ozer, 2001, Shin and Sohn, 2004) and neural network algorithms (Bloom, 2005, Chiu et al., 2009, Diez et al., 2008, Ha, 2007, Hsieh, 2004, Hung and Tsai, 2008, Kuo et al., 2006, Lee and Park, 2005, Potharst et al., 2001, Shin and Sohn, 2004) have been used in the related literature. Cluster-based segmentation methods, particularly hierarchical and non-hierarchical methods, have been widely used in the field. But, the hierarchical methods are criticised for non-recovery, while the non-hierarchical methods for their inability to initially determine the number of clusters (Lien, 2005). Hence, the integration of hierarchical and partitioning methods (two-stage methodology) is suggested to make the clustering results powerful for large databases (Kuo et al., 2002). None of those approaches, however, have the ability to establish non-strict customer segments that could play a significant role in today’s competitive consumer markets. Although there have been a few studies that utilised fuzzy segments they are not based on the effective two-stage methodology.

Complete Article List

Search this Journal:
Open Access Articles
Volume 11: 4 Issues (2020): 2 Released, 2 Forthcoming
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing