基于RF的中国贸易收支非线性回归模型研究  

An empirical study on balance of trade in China based on random forest regression model

作　　者：张仁寿[1] 罗林开[2] 吕伟航[2] 任晓怡[3] 

机构地区：[1]广州大学商学院,广东广州510006 [2]厦门大学自动化研究所,福建厦门361005 [3]广东金融学院,广东广州510520

出　　处：《广州大学学报（自然科学版）》2012年第4期94-97,共4页Journal of Guangzhou University:Natural Science Edition

基　　金：广州市教育局"高校社科"项目(63020)资助

摘　　要：研究了部分经济指标与贸易差额之间的非线性关系,该非线性模型采用随机森林回归模型,同时,得出了各个指标的影响因子.实验结果表明:我国的贸易差额主要受到M2,GDP,CPI,外汇储备,M1和工业总产值的影响,研究可供政府决策参考.This paper studies the nonlinear relation between the balance of trade and the affection factors. The nonlinear regression model for the balance of trade is constructed by random forest （RF） method. Moreover, the ranking of importance for affection factors is given out. The empirical results in this paper reveal the balance of trade in China is affected mainly by M2, GDP, CPI, Exchange reserves, M1 and Gross Industrial output, which is helpful for the government decision-making.

关 键 词：贸易差额 随机森林 回归模型 

分 类 号：F710[经济管理—产业经济]