Forecasting Needful Liquidity Automatic Teller Machines (ATM) With Linear Model (ARIMA) and Non Linear (Neural Networks)

The aim of dis study is forecasting needful liquidity automatic teller machines (ATM) With two linear model and non Linear. Providing financial resources at ATM machines, since it is important that is necessary to keep activity ATM to provide cash into applicant and security cash in machine. Results of Such analysis make it possible to can be necessary prediction for providing financial resources of ATM such as intelligent and maintain it active during peak demand hours. The statistical case of the research consisted of 7 ATMS of Economic Mehr Bank in Razavi Khorasan during the years 1386-1389. In this study, determining amount of cash taking from ATM using nonlinear method of artificial neural networks with multi-layer perception structure and error back propagation algorithm and linear method of ARIMA is examined to choose an optimal model. Demand of cash taking from date 1386/4 till 1389/7 (40 months) as training data and from the aban data of 1389 till the end of esfand 1389 is intended as Test data. The finding results indicate mastery artificial neural networks model to ARIMA model in predict needed cash of an ATM from economic mehr bank.