Gini impurity Machine Learning
noun phrase
Definition: A decision-tree impurity measure that quantifies how often a randomly chosen sample would be incorrectly labeled if it were labeled according to the class distribution in a node [scikit-learn Documentation].
Examples in context: “Important features of an RF model were selected using two approaches: standard scoring of features based on mean decrease in Gini impurity and application of SHAP’s TreeExplainer to RF.” [Chereda et al. 2024]
“In decision trees, the splitting rule (e.g., rules based on Gini impurity, variance reduction) is of great importance.” [Tyralis 2024]
Synonyms: gini index, gini
Related terms: impurity, information gain, entropy