Affiliations: Department of Industrial Engineering, Erciyes University, Kayseri, Turkey
Corresponding author: Mithat Zeydan, Department of Industrial Engineering, Erciyes University, Kayseri, Turkey. E-mail: [email protected].
Abstract: The solution of facility site selection problems is of utmost importance for the existence and future of the businesses. When planning for banking services, banks should ensure their accessibility standards by taking their customers and needs into account. For this reason, the accessibility of the services offered via ATM devices is very important. Contrary to the studies that have been widely done in the literature, in this study, the data mining techniques for determining locations of the ATMs, specifically decision tree algorithms were applied to select the best location among candidates considering new ATM installations. In practice, a data set of a private bank’s ATM’s, which in the boundaries of the districts connected to downtown of Kayseri, was used. A decision support model based on rule-based location detection was created from the data set discussed. The correct classification scales were evaluated by applying different algorithms for decision tree method. Among the decision tree algorithms, model estimation accuracy was found to be 92.96% on average. The results were obtained using STATISTICA 10.0 Software.
Keywords: Automatic Teller Machines (ATMs), decision tree algorithms, site selection