Presenting a new model for ATM demand scenario

Subject Areas : Futurology

Alireza Agha Gholizadeh Sayyar ¹ , Mohamadreza Motadel ² , Alireza Pour ebrahimi ³

1 - Ph.D. Student, Department of Industrial Management, Islamic Azad University, Science and Research branch, Tehran, Iran
2 - Assistant Professor, Department of Industrial Management, Islamic Azad University, Tehran Central Branch, Tehran, Iran [e-mail]
3 - Assistant Professor, Department of Industrial Management, Islamic Azad University, Karaj Branch, Karaj, Iran

Received: 2018-07-21 Accepted : 2019-01-20 Published : 2019-11-22

Keywords: Demand forecasting, K Means, data mining, Machine learning, C & R decision tree,

Abstract :

In today's competitive world, the ability to recognize predict customer demand is an important issue for the success of organizations. And since ATMs are one of the most important channels for cash distribution and one of the most fundamental criteria for assessing the level of service to banks,In this paper, the number of referrers to ATM devices is reviewed based on the timing and location of the devices. This article seeks to find a dynamic and functional model for predicting the number of referrers to each ATM depending on the time and location of the device. Hence, 378 ATM machines were used throughout the city of Tehran for a time period of one month, containing 69,418 records. Finally, with the help of clustering of statistical data in spatial and temporal dimensions, this model finally succeeds in learning the pattern in the macro data, and based on the decision tree, the predictor can predict the number of referents to each device, which after the algorithm is presented. In order to improve the quality of banking services and improve the performance of the ATM network, it is proposed to combine the optimal location of ATMs in spatial and temporal dimensions.