Effective multinational trade forecasting using LSTM recurrent neural network

作者：

Highlights：

• Changes in international trade constitute a crucial topic in economics.

• We developed a multivariate LSTM-based economic theory for trade forecasting.

• Trade data are considered as multivariate series with a two-way causal relationship.

• The multivariate LSTM-based economic theory demonstrates the superior performance.

摘要

•Changes in international trade constitute a crucial topic in economics.•We developed a multivariate LSTM-based economic theory for trade forecasting.•Trade data are considered as multivariate series with a two-way causal relationship.•The multivariate LSTM-based economic theory demonstrates the superior performance.

论文关键词：Deep learning,Long short-term memory,Trade forecasting,Multivariate time-series analysis,Recurrent neural network

论文评审过程：Received 1 September 2020, Revised 10 May 2021, Accepted 11 May 2021, Available online 18 May 2021, Version of Record 21 May 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115199