Python Machine Learning By Example

The metrics to measure a split

When selecting the best combination of features and values as the splitting point, two criteria, Gini impurity and information gain, can be used to measure the quality of separation.

Gini impurity as its name implies, measures the class impurity rate, the class mixture rate. For a dataset with K classes, suppose data from class k () takes up a fraction () of the entire dataset, the Gini impurity of such a dataset ...