This stats.stackexchange post contains explanation of how to interpret transformed variables in linear regression.
In particular, I found this snippet in Graham Cookson's answer (2nd answer):
Y and X -- a one unit increase in X would lead to a
increase/decrease in Y
Log Y and Log X -- a 1% increase in X would lead to a
% increase/decrease in Y
Log Y and X -- a one unit increase in X would lead to a ∗100
% increase/decrease in Y
Y and Log X -- a 1% increase in X would lead to a /100 increase/decrease in Y
My question is, what would the equivalent interpretations be when the context is a logistic regression and not a linear regression? Since you cannot transform the dependent variable in binary classification, it's really the last one that I'm interested in: Y and Log X -- a 1% increase in X would lead to a /100 increase/decrease in Y
If 2 input variables are log transformed and one has a resulting odds ratio 2 and one of 0.5, how can these be interpreted from a model explain-ability standpoint?
