For a logistic regression credit risk model, Weight of Evidence (WOE) encoding is applied to a high-cardinality categorical variable. Why is WOE particularly suited for logistic regression?
Select an answer to reveal the explanation.
Short Explanation
You're right because WOE turns each category into a log-odds value, so it lines up with the logit scale in logistic regression. That makes the encoded predictor directly compatible with the model.
Full Explanation
WOE encodes each category as log(event probability / non-event probability), directly expressing log-odds and creating a linear relationship with the logit outcome. WOE = log(Distribution of Events / Distribution of Non-Events) for each category, which maps to the logistic model's log-odds scale. This linearity makes the encoded predictor directly compatible with logistic regression. The option about reducing categories to one numeric column, saving computational cost, is not the primary statistical advantage. The option about assigning WOE 0 to unknown categories is not WOE's defining benefit. The option about SAS PROC LOGISTIC requiring WOE for CLASS variables with more than 10 levels is wrong; WOE is not required.