基于PCA-LASSO方法的行业市盈率预测及影响因素分析  

A Study on the Forecasting and Influencing Factors of P/E Based on PCA-LASSO Model

作　　者：孟洁莹 MENG Jieying(Zhejiang Post and Telecommunication College)

机构地区：[1]浙江邮电职业技术学院

出　　处：《中国商论》2021年第9期58-60,共3页China Journal of Commerce

摘　　要：本文基于分行业的横截面财务数据分析影响市盈率的主要因素,提出了PCA-LASSO模型及其精简模型方法,并对市盈率进行样本外预测,同时与传统的线性回归模型及LASSO回归模型的结果进行了比较。研究表明,在行业市盈率的样本外预测方面,所提出的PCA-LASSO模型及其精简模型方法明显优于已有的两种研究方法。所提模型方法融合了主成分回归和LASSO回归的优点,既完全消除了多重共线性又实现了对重要变量的选择,同时具有更高的预测精度,所提方法具有普遍适用性。Based on the industry’s cross-sectional financial data analysis of the main factors affecting the P/E ratio,PCALASSO model and its reduced model are proposed,and the P/E ratio is estimated out-of-sample and compared with the traditional linear regression model and LASSO regression model.The research shows that the proposed PCA-LASSO model and its reduced model method are obviously superior to the two existing research methods in the field of industry-specific earnings forecast.The proposed model incorporates the advantages of principal component regression and LASSO regression,which not only eliminates multicollinearity completely but also realizes the selection of important variables with higher prediction accuracy.The proposed method has universal applicability.

关 键 词：PCA-LASSO模型 市盈率 影响因素 样本外预测 

分 类 号：F832[经济管理—金融学]