Entropy Weighting Fuzzy Local Information C- means (EWFLICM) Clustering Approach for Proficient Customer Behaviour Mining in Telecom Industry

1C.K. Praseeda and Dr.B.L. Shivakumar


The data-intensive industries such as telecommunication organisations need business intelligence technique to improve the visibility and core operations of the organization. Customer clustering will enable to reliably analyse customer composition and improve service and marketing quality. Paying attention to customer service is regarded among the most vital components of the telecom industry's profitability. Clustering algorithms are often used for major data analysis technique for evaluating telecom customers. These algorithms are implemented to analyze customer’s behaviour patterns in the massive amount of telecom data. The current research, Customer Behaviour Mining (CBM) method is presented with fuzzy clustering algorithms. Considering the patterns of behaviour of customers and projecting how they can behave in the future, we develop a new methodology called Entropy weighting Fuzzy Local Information C- Means (EWFLICM) for clustering the large amount of data in telecom industries. Entropy function is introduced to EWFLICM clustering algorithm for weight updating. It optimizes the weighting parameter to control the contribution of the local factor for clustering of customers in mobile phone activity dataset. Five major activities such as received SMS, sent SMS, incoming calls, outgoing calls and internet activity have been carryout in mobile phone activity dataset. The cluster analysis is then carried out based on two factors, i.e. maximum period used by telecommunications services and the number of services used for each group of customers. Such support was committed to mining customer's future behaviour. The proposed EWFLICM system is described in terms of accuracy, Silhouette Coefficient (SC), Agglomerative Coefficient (AC), error rate and calculation time.


Customer Behaviour Mining (CBM), Telecom Industry, Fuzzy C Means (FCM), Fuzzy Local Information C- Means (FLCIM), and Entropy Weighting Fuzzy Local Information C- Means (EWFLICM) Algorithm.

Paper Details
IssueIssue 4