计算机与现代化

• 算法设计与分析 • 上一篇    下一篇

基于PSO优化混合RVM模型的进出口贸易预测算法

  

  1. (天津大学管理与经济学部,天津300072)
  • 收稿日期:2014-04-16 出版日期:2014-08-15 发布日期:2014-08-19
  • 作者简介:白霜(1981-),男,吉林敦化人,天津大学管理与经济学部硕士研究生,研究方向:进出口贸易预测。

Import and Export Trade Prediction Algorithm Based on #br# Multi-kernel Weighting RVM Model Optimized by PSO

  1. (College of Management and Economics, Tianjin University, Tianjin 300072, China)
  • Received:2014-04-16 Online:2014-08-15 Published:2014-08-19

摘要:

由于进出口贸易额波动较大,影响因素较多,一般预测算法难以得到较为准确的预测结果。针对该问题,提出基于PSO优化混合
RVM模型的贸易预测方法。该方法首先找出影响进出口贸易的指标并通过主成分分析方法提取出指标的主因子作为模型的输入数据。然后
在多个不同核函数的单一核RVM模型训练的基础上,根据单一核RVM模型预测误差采用多核加权的方法构建混合核RVM模型,最后通过PSO
优化混合核模型参数以提高预测准确性。以深圳进出口贸易预测为例验证该方法能够较为准确地预测进出口贸易值。

关键词: 贸易预测, 关联向量机, 粒子群优化, 混合核

Abstract:

Because of high fluctuation and many influence factors, common algorithm cannot get precise predictions.
According to the problem, the article presents a new import and export trade prediction algorithm based on multi-
kernel RVM model optimized by PSO. Firstly, the article finds factors influencing import and export trade and
extracting main factors with PCA. Then on the basis of training of several single kernel RVM models based on
different kernel functions, multi-kernel RVM model is built according to the prediction error. Lastly, parameters
of multi-kernel RVM model are optimized by PSO to heighten precision of prediction. The emulation experiment shows
that this method can gain more accurate prediction result in the prediction of Shenzhen import and export trade.

Key words: import and export trade prediction, RVM, PSO optimization, multi-kernel

中图分类号: