Forecasting of e-commerce transaction volume using a hybrid of extreme learning machine and improved moth-flame optimization algorithm

Abstract

The rapid development of e-commerce has resulted in optimization of the industrial structure of Chinese enterprises and has improved the Chinese economy. E-commerce transaction volume is an evaluation index used to determine the development level of e-commerce. This study proposed a model for forecasting e-commerce transaction volume. First, a hybrid moth–flame optimization algorithm (HMFO) was proposed. The convergence ability of the HMFO algorithm was analyzed on the basis of test functions. Second, using data provided by the China Internet Network Information Center, factors influencing e-commerce transaction volume were analyzed. The input variables of the e-commerce transaction volume prediction model were selected by analyzing correlation coefficients. Finally, a hybrid extreme learning machine and hybrid-strategy-based HMFO (ELM-HMFO) method was proposed to predict the volume of e-commerce transactions. The prediction results revealed that the root mean square error of the proposed ELM-HMFO model was smaller than 0.5, and the determination coefficient was 0.99, which indicated that the forecast e-commerce transaction volume was satisfactory. The proposed ELM-HMFO model can promote the industrial upgrading and development of e-commerce in China.