This question already has an answer here:

What is the implication if I don't fix a logistic regression that has complete or quasi separation? can I still read the marginal effects or are they not going to be valid?

My exercise is actually to just find out which independent variables are most predictive of Y.

I read some responses to complete/quasi-separation and I tried using logistf package for R but got this error message "NA/NaN/Inf in foreign function call logistf". why does this arises?


marked as duplicate by kjetil b halvorsen, Peter Flom regression Sep 5 '17 at 22:39

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.