header image
Home
Why PLS sometimes has more than one component? PDF Print E-mail
Written by Johan Trygg   
I want to discuss some algebraic aspects of the popular OLS and PLS regression methods in chemometrics taken from my thesis (2001). The main point I want to make here is that single-y PLS models should only have one (predictive) component.
If it has more than one, it means that there is strong systematic variation in X irrelevant for the prediction of Y. Due to that, the interpretation, not the prediction, of the PLS model will suffer in direct relation to the number of additional PLS components. Let me explain what I mean...
 
< Prev
Search website

Editorial flash
Can we use the regression coefficient profile for selecting PLS components?
The linear model y=Xb + f is often used in chemometrics. Unfortunately, we have put too much emphasis on the importance of the regression coefficient profile b, and mistakenly assumed or believed that it should be a good estimate of the "pure profile". This is not true, and I will argue why in later editorials. But this time, I want to address an earlier question that was posed in relation to the regression coefficient profile.
Read more...
News flash
Discussion Forum is available again

Dear Chemometrics.se visitor,

Due to the many requests to reopen the Chemometrics Discussion Forum, I have made it available here at Chemometrics.se (under Discussion Forum in the menu).

Read more...
Tutorial flash
The Use of Virtual Calibrations to Facilitate Understanding of Factor Analysis

Factor Analysis (FA), which includes Principal Component Analysis (PCA) and Partial Least Squares (PLS), is more and more employed in academia and industry for various purposes such as spectrometer calibration, process modeling, data mining, quality control, etc. While software offering friendly interfaces have contributed to make this approach extremely popular, FA remains far from being straightforward, and examples of inappropriate use are not rare.

Read more...