Potential Indicators Based Neural Networks for Cash Forecasting of an ATM

摘要: Cash forecasting is one of the important tasks in domain computational finance. A number tools have been developed by various groups researchers and are being used banks or corporate to identify future cash needs. However, due high degree non-linearity problem surrounded many local optimal solutions, this paper propose a multi-layer locally tuned perceptron (MLTP) forecast needs at same time reduce users frustration. It uses fine MLTP daily demand an automated teller machine (ATM). Further, potential indicators making model robust terms its efficiency accuracy. The accuracy compared against traditional series method. Furthermore, it validated using past data collected from SBI ATM Bhadrak district Odisha, India. performance method encouraging. This system can be scaled for all branches bank area incorporating historical these branches.