Hey, #RStats and other #statistics folks. Let's say I have a very large number of highly correlated variables and I want to find a series of linear combinations of those variables that are maximally correlated with a target measurement. What method would you use to find them?

Sparse Partial Least Squares sounds like what I need, but I get the strange results that all linear combinations are almost identical.

Follow

@eliocamp another thing to check is canonical correlation analysis

@dimokaramanlis I tried that. I might be missing something but the R function I used only returned one set of coefficients. Is that wrong?

@eliocamp Not sure about the implementation of R, but probably these coefficients are the linear combination weights that lead to a variable that maximally correlates with the measurement you want to predict. Aren't those what you are looking for?

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.