Adaptive Automated Teller Machines — Part I

During the past few years, the banking sector has started providing a variety of services to its customers. One of the most significant of such services has been the introduction of the Automated Teller Machines (ATMs) for providing online support to bank customers. The use of ATMs has reached its zenith in every developed country, and thousands of ATM transactions are occurring on a daily basis. In order to increase the customers' satisfaction and to provide them with more user-friendly ATM interfaces, it becomes important to mine the ATM transactions to discover useful patterns about the customers' interacting behaviors. In this work, we apply diverse data mining techniques to an ATM transaction dataset obtained from an international bank based in the Middle East. We pre-process this dataset, and convert it into a specific XML format, called MXML, in order to mine it through the ProM (process mining) tool. We divide our work into two papers, i.e. Part I and Part II. In Part I (this paper), we present the background knowledge and functionality related to the pre-processing of ATM dataset, and its conversion to MXML, along with the related work. Then, in Part II (companion paper), we present our results related to the data mining of the ATM dataset, e.g., the amount withdrawal distribution of the ATM customers, based on time and location of the ATM terminals. Based on these mining outputs, we are currently developing an adaptive ATM interface which caters for the specific preferences of ATM users, e.g., by showing different GUIs at different time intervals.