基于因子-主成分回归分析的股价技术分析研究

Factor-Principal Component Regression Analysis on Technical Analysis of Stock

介绍了回归分析中多元线性回归的理论及应用方法,并以股价技术指标为研究对象,利用spss统计分析软件,建立了短期股价变动的多元线性回归模型。同时讨论了被选为自变量的参数之间存在的多重共线性问题,并分析该问题对线性回归分析结果造成的影响。因子-主成分分析的核心是用较少的相互独立的因子反映原有变量的绝大部分信息。主成分分析的主要思想是:从自变量中提取出新的变量,这些变量是原变量的适当线性组合,并且互不相关,因此应用SPSS软件进行数据缩减、提取主成分,并以主成分因子为新的自变量建立主成分回归方程,消除了多重共线性对回归模型的影响。最后对不同模型的测试结果进行了比较、分析,验证了因子-主成分分析在解决实际经济问题中的有效性。

This paper describes the theory and the method of application of the multiple linear regression analysis.We set up the regression model on the variation of the stock-prices with the specifications by using the spss software,and discuss the effect of multicollinearity.Then,we discuss the multicol linearity of the factors which we choose to set up the model and the effect of multicollinearity. Factor-principal regression is the method that has less factors which are independent and can reflect the most information of the original factors chosen,and these new factors are original factors＇ linear combination,so we use spss software to choose the principal factors,and set up the new model which solves the multicollinearity with them.At the end of this paper,we compare the results of the two different models,so the validity of the factor-principal regression analysis has been confirmed.