A Study of Partial Least Squares (Case Study: Cox-PLS Regression)

Abstract

Indication of multicollinearity in regression analysis will lead to wrong interpretation when interpreting the results. One of the handling of the case of multicollinearity is to use of PLS (partial least squares). The purpose of this study is to provide a general overview of PLS. The results of this study are in general PLS study along both the concept and the classification methodology and its application. PLS is generally divided into two branches, namely PLS regression and path analysis. In the application of PLS, the data used TB patient survival (tubercolosis) in Yogyakarta, which is obtained from a private hospital in Yogyakarta. The data were analyzed using Cox regression, but there is multicollinearity so then there is an error in the interpretation of the significance of the model. By using PLS-Cox regression, we obtained one PLS component consisting of one independent variable, namely class care.

Key words: PLS regression, PLS path modelling, Cox regression