hooman razavi - pulseware corporation
bahareh hassanpour - ieee student member
ahmad karimisefat - phd.ocs company
jean fabrice lebraty - professor at universite de lyon ۳ &laboratoire magellan

One of the main indicators for making decision in banking area is the income of ATMs. Banks are tended to increase their acquirer incomes against the costs of an ATM which include installation, setting up and maintenance. Due to dependence of cash income on various factors such as geography and physical, ATM’s income function is complex which having recognition of its behavior could leads to discovering the mathematical relation between features of an ATM device and the weights of them versus ATM’s income. In this study based on artificial neural networks (ANNs) prediction model and the real data of some ATMs, a comprehensive income model presented which can approximate the income of assumptive ATM. In order to have accurate analysis and better training for ANNs, artificial methods are used to find out the correlation between ATM features and its income and also eliminate uncorrelated features. Applying these analyses will help banking organizations and institutions with making future decisions and also finding locations, situations and opportunities for presentation of ATM services to customers

ATM Income, Statistical Methods, Data Mining, Artificial Neural networks, Location, Correlation