数学理论与应用 ›› 2020, Vol. 40 ›› Issue (3): 101-109.
• • 上一篇 下一篇
余奇迪 戴家佳*
贵州大学数学与统计学院、贵州省博弈决策与控制系统重点实验室,贵州 贵阳 550025
出版日期:
发布日期:
通讯作者:
作者简介:
基金资助:
贵州省数据驱动建模学习与优化创新团队
Online:
Published:
摘要:
本文基于中国市场3465家上市公司7年的数据,首先利用随机森林算法提取出43个因子,并对选取出的43个因子再利用lasso方法进行特征选取,最后选取出11个重要因子。然后采用logistic回归构建第一种预测模型,之后利用决策树模型构建第二种预测模型,最后基于损失函数确定权重的组合模型将第一种预测模型与第二种预测模型进行线性组合建立组合模型。实证结果表明,基于组合模型的预测准确率相比单一模型提高了1.39%。
关键词: 随机森林, logistic回归, 决策树, 组合模型, 高送转
随机森林,
Abstract:
Based on the data of 3465 listed companies in the Chinese market in 7 years,this paper firstly extracted 43 factors by using random forest algorithm,and then used Lasso method to select the characteristics of the 43 factors selected,and finally selected 11 important factors.Then logistic regression is used to build the first prediction model,and then the decision tree model is used to build the second prediction model. Finally,the combination model based on the loss function to determine the weight is linear combination of the first prediction model and the second prediction model to build the combination model. The empirical results show that the prediction accuracy of the combined model is 1.39% higher than that of the single model.
Key words: random , forest, Logistic , regression, Decision , tree, Combinatorial , model, High , to , turn
random ,
余奇迪 戴家佳. 基于组合模型的上市公司高送转预测[J]. 数学理论与应用, 2020, 40(3): 101-109.
0 / / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: https://mta.csu.edu.cn/CN/
https://mta.csu.edu.cn/CN/Y2020/V40/I3/101