Enum linfa_trees::SplitQuality

pub enum SplitQuality {
    Gini,
    Entropy,
}

The metric used to determine the feature by which a node is split

Variants

Gini

Measures the degree of probability of a randomly chosen point in the subtree being misclassified, defined as one minus the sum over all labels of the squared probability of encountering that label. The Gini index of the root is given by the weighted sum of the indexes of its two subtrees. At each step the split is applied to the feature which decreases the most the Gini impurity of the root.

Entropy

Measures the entropy of a subtree, defined as the sum over all labels of the probability of encountering that label in the subtree times its logarithm in base two, with negative sign. The entropy of the root minus the weighted sum of the entropy of its two subtrees defines the “information gain” obtained by applying the split. At each step the split is applied to the feature with the biggest information gain

Trait Implementations

impl Clone for SplitQuality

fn clone(&self) -> SplitQuality

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SplitQuality

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for SplitQuality