摘要
基于中国经济信息网发布的CCI指数,结合DEGWO差分算法和BP神经网络回归,构建了机器学习方式下的DEGWO-BP合成算法,从而对消费者信心指数进行预测和拟合;实证结果显示,采用DEGWO-BP算法的消费者信心指数的累积误差最低仅为43.4631;模型的平均绝对误差最低,模型具备最小的偏差水平;模型的极端偏差值最小,模型具备最强的稳定度。
On the basis of the cci index published by China Economic Information Network,combined with DEGWO difference algorithm and BP neural network regression, the DEGWO-BP synthesis algorithm under machine learning mode is constructed to predict and fit the consumer confidence index. Empirical results show that the cumulative error of consumer confidence index using DEGWO-BP algorithm is only 43.4631. The average absolute error of the model is the lowest and the model has the minimum deviation level. The extreme deviation value of the model is the smallest, and the model has the strongest stability.
作者
杨学分
Yang Xuefen(School of Economics and Trade Management,Anhui National Defense Science and Technology Vocational College,Liu’an Anhui 273001)
出处
《保山学院学报》
2020年第1期92-97,共6页
JOURNAL OF BAOSHAN UNIVERSITY
基金
安徽省教育厅2016年度人文社会科学研究重点项目:“‘互联网+’背景下皖西特色农产品电子商务发展路经”(项目编号:SK2016A0203)
安徽省教育厅2018年度人文社会科学研究重点项目:“‘互联网+’背景下产业园区品牌建设的路经研究——以六安市承接产业转移集中示范园区为例”(项目编号:SK2018AD924)。
