Fuzzy Time Series Customers Prediction: Case Study of an E-Commerce Cash Flow Service Provider

Abstract

With lower operational costs, many small and medium-sized enterprises (SMEs) trade via e-commerce, but without the abilities to develop the expensive payment system. Therefore, a cash flow service provider plays critical roles to complete the online transactions. A cash flow service provider must precisely predict those outsourcing customers to serve the customers well under the correctly prepared facilities. Since the adaptive neuro-fuzzy inference systems (ANFIS) have demonstrated prediction efficiency for fuzzy circumstances in many fields, this study attempts to innovate deploying the ANFIS model on the time series predictions for e-commerce cash flow service customers. Moreover, this study takes an e-commerce cash flow service provider in Taiwan for numerical analysis. For the ANFIS predictions, an acceptable prediction error rate of 5.6% is achieved. The results show that fashion industry tops the highest customers share for outsourcing the cash flow services; and the credit cards top the highest share in the payment media choices.