Bootstrap Confidence Interval for Regression Coefficients of PCR  
Author Jieun Kim

 
Co-Author(s) Seohoon Jin

 
Abstract There is a problem of multicollinearity arising from the correlation between independent variables in multiple regression analysis. Principal components regression (PCR) analysis is one of the methods used to solve this problem. Since PCR uses the principal component as an independent variable, it is difficult to identify the significance of the original variables. We used the bootstrap to build confidence interval of regression coefficients of the original variables and interpreted the meaning of the variables.

 
Keywords PC(principal components), PCR(principal components regression), bootstrap, confidence interval
   
    Article #:  DSBFI19-16
 
Proceedings of ISSAT International Conference on Data Science in Business, Finance and Industry
July 3-5, 2019 - Da Nang, Vietnam