Ordinal Logistic Regression
Train Ordinal Logistic Regression to predict categorical outcomes
Specialized for predicting ordered categories where the order matters (like ratings: poor < fair < good < excellent).
When to use:
- Target has natural ordering
- Categories are not arbitrary (unlike regular classification)
- Examples: satisfaction ratings, education levels, disease stages
Strengths: Respects category order, more accurate than treating ordinal as nominal Weaknesses: Requires ordered target, assumes proportional odds
Model Parameters
Same parameters as Logistic Regression, but respects the ordering of classes.