Preporato

Trees and ensembles

Splits, impurity and boosting: how tabular models really work.

Start 5 problems

Problems

  1. 1Gini impurity of a nodeEasy · 7 min · 11 hidden testsSolution free
  2. 2Best split on one featureMedium · 15 min · 12 hidden tests
  3. 3Information gain of a splitMedium · 12 min · 11 hidden tests
  4. 4Majority vote of a forestEasy · 8 min · 10 hidden tests
  5. 5One gradient boosting roundMedium · 14 min · 10 hidden tests

Practice it at full scale

Hands-on labs on real tooling, in your browser.

Gradient Boosting with XGBoostExplain a Model