PCA on nominal variables

I am a bit confused if nominal (discrete and boolean) variables can be used for PCA (in meaningful way).
I read through many articles, forums, but unfortunately I am not qualified enough to decide between the different and sometimes contradictory answers/statements. Unfortunately, many real life cases are containing mixed numerical and nominal variables.

And this questions applies to the dummy variables as the outcome of the step_dummy() (as seen in a recent tidymodels tutorial).

This question is haunting me. Can nominal variables used in k-means clustering? ...

They can so long as they are normalized to the same scale as the other predictors.

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.