Customer segmentation in intelligent learning mechanism in E- banking using Frequent Item-set Hierarchical Clustering

Abstract

Direct marketing is a business model that uses data mining techniques and marketing databases for personalization and business intelligence. It is a new approach that uses interactive one-to-one communication between marketer and customer aimed at specific customers with personalized advertising and promotional campaigns. There are few attempts of using Data Mining for Direct Marketing. The major problems encountered are mining of the huge volume of customer topographies for marketing purpose, churn management and deficiency of binary classification algorithms. In this paper, we propose a novel and efficient learning algorithm called CSBC for intelligent Learning based Direct Marketing using Frequent Item-set Hierarchical Clustering which can be used by the marketers to personalize the next campaign. Frequent-Itemset Hierarchical clustering and ranking method is incorporated in CSBC to grade/segment the customers into more than two classes. A classifier produced rules would be used to predict the new customer data. Also, the marketers will be able to predict the churners by evaluating the predicted classes/ranks along with the relationship on the client information attributes of dataset. Prediction of respondents and non-respondents from new data which will be useful for direct marketing is done using fully automated knowledge extraction procedure. Customer Segmentation is evaluated using lift measure and ROC curve measure in experiments.