International Journal of Management, Accounting and Economics
Hits: 710 Times
Downloads: 713 Times
Volume 3, No. 8, August 2016 Pages: 534 - 543
Classification of Bank Customers by Data Mining: a Case Study of Mellat Bank branches in Shiraz
Dariush Farid, Hojjatollah Sadeghi, Elahe Hajigol, Nadiya Zarmehr Parirooy
Corresponding author:
nadiya[dot]zarmehr[at]yahoo[dot]com
Abstract:
This research predicts through studying significant factors in customer relationship management and applying data mining in bank. Financial institutions and other firms in competitive market need to follow proper understanding of customer behavior. Customers’ data are analyzed to identify specific opportunities and investment, to classify and predict the behaviors; further, data are eventually used for decision-making. Therefore, data mining as knowledge exploring (discovery) approach plays a significant role through a variety of algorithms. This study classifies bank customers by using decision tree algorithm. Three decision tree models including ID3, C4.5, and CART were applied for classifying and finally for prediction. Results of simple sampling method and k-fold cross validation show that forecast accuracy of C4.5 decision tree using simple sampling was higher than other models. Thus, predicting customers’ behavior through C4.5 decision tree was considered the ideal prediction for bank. This research predicts through studying significant factors in customer relationship management and applying data mining in bank. Financial institutions and other firms in competitive market need to follow proper understanding of customer behavior. Customers’ data are analyzed to identify specific opportunities and investment, to classify and predict the behaviors; further, data are eventually used for decision-making. Therefore, data mining as knowledge exploring (discovery) approach plays a significant role through a variety of algorithms. This study classifies bank customers by using decision tree algorithm. Three decision tree models including ID3, C4.5, and CART were applied for classifying and finally for prediction. Results of simple sampling method and k-fold cross validation show that forecast accuracy of C4.5 decision tree using simple sampling was higher than other models. Thus, predicting customers’ behavior through C4.5 decision tree was considered the ideal prediction for bank.
Keywords:
Validation, data mining, decision tree, customer relationship management.
Full Text (PDF)