摘要
新冠肺炎疫情的爆发对中国经济产生了深远影响,在后疫情时代对未来短期的进出口额分析与预测将对我国制定政策、企业规划生产计划都具有十分重要的意义。本文将传统的Prophet、LSTM、ARIMA三种模型进行加权融合,并对这三种模型的权重进行统计估计,提出了一种进出口额预测分析的PLA (Prophet-LSTM-ARIMA)融合模型。最后,对2011年1月至2022年3月共135个月的真实进出口数据进行描述性统计分析和统计建模分析,预测了2022年4月至2023年3月的进出口值,并对各模型进行性能对比分析。实验结果表明:本文提出的PLA融合模型对各大类进出口商品预测的平均相对误差率比传统的Prophet、LSTM、ARIMA模型小,具有更好的预测效果,该模型对进出口贸易的科学决策提供了一定的理论支撑和实际指导。
The outbreak of the new crown pneumonia epidemic has had a profound impact on China’s economy. In the post-epidemic era, the analysis and forecast of short-term import and export volume will be of great significance for my country to formulate policies and plan production plans for enterprises. This paper proposes a PLA (Prophet-LSTM-ARIMA) fusion model for the forecast analysis of import and export value, which combines the traditional three models of Prophet, LSTM and ARIMA by weight, and performs statistical estimation on the weights of these three models. Finally, descrip-tive statistical analysis and statistical modeling analysis of real import and export data were con-ducted for a total of 135 months from January 2011 to March 2022, and forecast import and export value from April 2022 to March 2023, moreover, perform the performance comparison analysis of each model. The experimental results show that the average relative error rate of the PLA fusion model proposed in this paper is smaller than that of the traditional Prophet, LSTM, and ARIMA models for the prediction of various types of import and export commodities, and provides certain theoretical support and practical guidance for scientific decision-making of import and export trade.
出处
《应用数学进展》
2022年第10期7291-7301,共11页
Advances in Applied Mathematics