Business Management Dynamics

Double Blind Peer Reviewed - Open Access Jounal

Home  |   Contact Us

 

ISSN: 2047-7031

bmd Business and Management Dynamics bmd
ISSN: 2047-7031  
Volume     Issue     
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Article Abstract
Ant Colony Algorithm Compared with Other Data Mining Algorithms for Segmenting Bank Credit Customers
Keywords:  Credit customers, Data mining, Credit risk, Ant colony
Ali Reza Poor Ebrahimi and Hossein Farzad
This research aims to segment bank's credit customers in terms of received facilities repayment risk. In this research, using corporate customers' data, which are available in information systems of the studied bank credits department, data mining process is carried out. At first, the required data were collected and pre-processing was carried out on them. Influencing variables in model were identified, and ants' colony algorithm was run on the final data, then the results of this algorithm were compared with some of the other algorithm of the category. The obtained results showed that ants' colony algorithm, compared to traditional trees of CART, CHAID, and QUEST, neural networks, distinctive analysis, logistic regression, and Bayesian networks, has better results in terms of the exactness of customers' separation and segmentation, but it is less accurate compared to tree models of C4.5, and supporting vector machine (SVM). On the other hand, ants' colony technique has better prediction power, compared to prediction performance of the bank's validator experts. Finally, a model which has the highest accuracy in predicting the customers' category was recommended as the proposed algorithm. It is noteworthy that in this research, Ant miner software has been used to run ants' colony algorithm.
01-07   |  View PDF