Presentation of customer foreclosure model of e-banking services using decision tree and data mining method

Abstract

In this research, we use the methods of data mining to discover the pattern of customer reversal of banking services. The statistical community of the Ayandeh bank customers who have been active in the provision of electronic services, among which 149 people were selected by sampling method in West Tehran's branches. In this research, three tree decision algorithms for decision tree C & R TREE, QUEST TREE and CHILD TREE are used to predict the deviations, reversal patterns and the most important attributes that affect it. Based on the results of decision tree algorithms, these algorithms with acceptable accuracy (above 80%) can predict customer behavior. Based on the results, the C & R Tree decision tree algorithm, better than other algorithms, can turn customers away Predict Also, based on decision trees, and considering the percentage of deviations in each of the nodes, it is possible to discover the laws that lead to customer rejection. In this regard, it is recommended that these rules be applied in marketing and customer retention guidelines. Also, based on the results of this study, five important attributes for predicting customer rejection are, respectively, occupation, branch grade, education, inventory grade and type of investment that banks should pay particular attention to.