Ordinal Logistic Regression
Logistic regression for ordered/ranked categorical outcomes
Ordinal Logistic Regression extends standard logistic regression to handle target variables with a natural ordering (e.g., low/medium/high, star ratings), respecting the order between categories during training.
When to use:
- Classification where categories have a meaningful order (ratings, severity levels, grades)
- When treating ordered categories as nominal classes would discard useful information
- Survey responses or Likert-scale targets
Input: Tabular data with the feature columns defined during training Output: Predicted ordinal class label and class probabilities
Model Settings (set during training, used at inference)
Penalty (default: l2) Regularization type applied to model coefficients.
C (default: 1.0) Inverse regularization strength. Lower values increase regularization.
Solver (default: lbfgs) Optimization algorithm used during training.
Max Iterations (default: 100) Maximum solver iterations.
Class Weight (default: null)
Set to balanced for imbalanced ordinal class distributions.
Inference Settings
No dedicated inference-time settings. The ordered class structure and thresholds are determined by the trained model.