How to retain factor scores in PLS

[ChuksManuel]

Hello programmers,

I am running a PLS analysis for a diet data and i want to know how i can retain the factors scores in PLS.

I am very familiar with retaining factor scores of the PCA analysis for use in my regression model but how to retain factors that are derived from PLS is a little problematic. Attached is my pls code.

proc pls data = zthree method = pls  nfac=4 varss details;
  model iron calc = T_mean l_mean vc_mean vd_mean vr_mean cn_mean e_mean mp_mean mr_mean /solution;
  output out = pattern xscore = scorex yscore = scorey;
run;

[Reeza]

Here's some instructions and explanations on how to capture output that is shown.
https://blogs.sas.com/content/sastraining/2017/03/31/capturing-output-from-any-procedure-with-an-ods...

---

@ChuksManuel wrote:

Hello programmers,

I am running a PLS analysis for a diet data and i want to know how i can retain the factors scores in PLS.

I am very familiar with retaining factor scores of the PCA analysis for use in my regression model but how to retain factors that are derived from PLS is a little problematic. Attached is my pls code.

proc pls data = zthree method = pls  nfac=4 varss details;
  model iron calc = T_mean l_mean vc_mean vd_mean vr_mean cn_mean e_mean mp_mean mr_mean /solution;
  output out = pattern xscore = scorex yscore = scorey;
run;

---

[PaigeMiller · Accepted solution]

output out = pattern xscore = scorex yscore = scorey;

Your code captures the scores into a SAS data set, does it not?

[ChuksManuel]

Thank you!

It so happens that the scores were actually retained and the scores that i needed for my regression model was the XScores.

[PaigeMiller]

---

@ChuksManuel wrote:

Thank you!

It so happens that the scores were actually retained and the scores that i needed for my regression model was the XScores.

---

I'm curious, as you wouldn't normally use the X-Scores from PLS in a regression model; they are already computed as part of a regression model, and using them in another regression model doesn't seem to make sense here.

[ChuksManuel]

In the PCA, i retained the rotated factor patterns and used them for regression.

In the PLC, is there a way i can retain the derived factors for use in a regression analysis?

[PaigeMiller]

In PLS the scores are already part of a regression. PLS is a regression.

Your terminology has changed from "scores" to "derived factors", not the same thing at all. Which part of PLS are you talking about?

[ChuksManuel]

I figured it out. Thanks for the help.

[Ksharp]

Yeah. I also have the same requirement, which I can deploy PLS into other language like Java or Python .

[Reeza]

---

@Ksharp wrote:

Yeah. I also have the same requirement, which I can deploy PLS into other language like Java or Python .

---

If you have EM you can get the PMML code which Java or Python should consume. 

[Ksharp]

Good news.

PLS model is available for PMML code ?

[Reeza]

I don't know that one....but what happens if you build a model that has a code node and generate the PMML from that? I'd test it and see what happens if you have EM but unfortunately I don't have access to EM to test anything 😞