EasyDeepLearn

What are monotonic constraints in gradient boosting and when are they worth using?

hard

Answer

  • They force the predicted output to move in one direction as a feature increases.
  • Useful when domain knowledge is certain: risk should not fall as debt rises, and price should not fall as square footage rises.
  • Benefits are regulatory defensibility, more stable behaviour in sparse regions of the feature space, and mild regularization that often helps on small data.
  • Cost is bias if the true relationship is not monotone, so check that the constraint does not degrade validation performance.
  • Supported directly in XGBoost, LightGBM and CatBoost.
Check yourself — multiple choice
  • They speed up training
  • They force the prediction to move one way with a feature — good for defensibility and stability, biased if the truth is not monotone
  • They only work in linear models
  • They remove the need for validation

Monotonic constraints encode certain domain knowledge, trading potential bias for stability.

#gradient-boosting#regularization

Practise Supervised Learning

215 interview questions in this topic.

Related questions