Comparison of Principal Component Regression and Partial Least Squares Regression by R
Yin Xiaocui
Abstract
Yin Xiaocui
Abstract
Principle component regression (PCR) and partial least squares regression (PLSR) are two methodologies commonly used to solve dimension reduction and the multi-co linearity problem. The goal of this paper is to analyze and predict dependent variable from predictor variables using the methods of PCR and PLSR by an example. The results show that the main difference between PCR and PLSR is that PCR often needs more components than PLSR to achieve the same prediction error and prediction ability.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Principle component regression (PCR) and partial least squares regression (PLSR) are two methodologies commonly used to solve dimension reduction and the multi-co linearity problem. The goal of this paper is to analyze and predict dependent variable from predictor variables using the methods of PCR and PLSR by an example. The results show that the main difference between PCR and PLSR is that PCR often needs more components than PLSR to achieve the same prediction error and prediction ability.
Key concepts: Partial least squares regression, Principal component regression, Statistics, Principal component analysis, Regression analysis, Regression, Mathematics, Dimension (graph theory)