Proc PLS, NIPALS and missing values

[deleted_user]

Hello,
I would like to make a PLS regression with the Proc PLS. I have some missing values in my dataset but I thought that the NIPALS algorithm could run with it. By default, it’s not true. There is a MISSING option in the procedure: with MISSING=NONE, observations with any missing variables are excluded from the analysis; with the two others options, AVG and EM, missing values are replaced by estimations. I think that is contradictory with what the NIPALS algorithm allows. Could you correct me if I make a mistake? Otherwise, how is it possible to run the Proc PLS with the NIPALS algorithm and missing values?

Thanks.

rcas


PS : Sorry for my english.

[Paige]

You are right that in theory, the NIPALS algorithm can be run with missing values for some data.

However, I believe that the PROC PLS implementation of NIPALS doesn't allow for missing values. You have to use MISSING=EM or MISSING=AVG, which as you said, fills in estimated values.

[deleted_user]

Thanks for the answer.