数据驱动模式下基于非平行超平面SVM算法的贸易经济预测
Data-driven Prediction of Trade Economy Based on Non-parallel Hyperplane SVM Algorithm
巢瑞云 1徐健2
作者信息
- 1. 广州南洋理工职业学院 经济管理学院,广州 510900
- 2. 桂林医学院 智能医学与生物技术学院,广西 桂林 541199
- 折叠
摘要
数据驱动的多元化发展导致数据异构性增强、维度提升和特征量规模扩大,给贸易经济分析带来更大挑战.为了提高贸易经济分析的科学性,采用非平行超平面支持向量机算法(support vector machine,SVM)对贸易经济进行预测分析.首先,根据贸易经济影响因素进行主成分分析,获取影响贸易经济的关键特征,并对特征进行量化和去噪处理.然后,采用广义特征值最接近支持向量机(proximal support vector machine via generalized eigenvalues,GEPSVM)进行贸易经济预测分类.根据预测指标要求,选择核函数GEPSVM算法(KGEPSVM算法)对分类的非平行超平面求解,通过类别划分函数获得经济预测结果.实证分析表明,对比常用的非平行超平面支持向量机算法,所提算法的贸易经济预测性能更优,而且在常用贸易经济指标的预测中,表现出较高预测精度和稳定性.
Abstract
Data-driven diversification leads to the enhancement of data heterogeneity,the dimension promotion and the increase of feature quantity,which poses greater challenge to trade and economic analysis.To achieve more scientific analysis of trade economy,nonparallel hyperplane support vector machine(SVM)algorithm is used to analyze and predict trade economy.Firstly,principal components are analyzed to obtain the key features affecting trade economy,and quantify and denoise these features.Then,the Proximal Support Vector Machine via Generalized Eigenvalues(GEPSVM)is used to classify the trade economic prediction.According to the requirements of prediction index,the kernel function GEPSVM algorithm(KGEPSVM algorithm)is adopted to solve the nonparallel hyperplane of classification,and the economic prediction results are obtained by using classification function.The empirical study shows that compared with the commonly used nonparallel hyperplane support vector machine algorithm,the proposed algorithm has better performance with higher accuracy and stability in prediction.
关键词
贸易经济预测/数据经济/非平行超平面/支持向量机/KGEPSVM算法/算法比较Key words
trade economic forecast/data economy/non-parallel hyperplane/support vector machine/KGEPSV algorithm/algorithm comparison引用本文复制引用
基金项目
2021年广东省普通高校特色新型智库项目(2021TSZK021)
出版年
2024
国家科技期刊平台
NSTL
万方数据