Module classifier 

Classification tree learners.

The module intentionally supports multiple tree families because they express different tradeoffs:

• id3 / c45 keep multiway splits for categorical-like binned features.
• cart is the standard binary threshold learner.
• randomized keeps the CART structure but cheapens split search.
• oblivious uses one split per depth, which is attractive for some runtime layouts and boosting-style ensembles.

The hot numeric paths are written around binned histograms and in-place row partitioning. That is why many helpers operate on row-index buffers instead of allocating fresh row vectors at every recursive step.

Structs

DecisionTreeClassifier

Concrete trained classification tree.

DecisionTreeOptions

Shared training controls for classification tree learners.

Enums

DecisionTreeAlgorithm

DecisionTreeError

Functions

train_c45

train_c45_with_criterion

train_cart

train_cart_with_criterion

train_id3

train_id3_with_criterion

train_oblivious

train_oblivious_with_criterion

train_randomized

train_randomized_with_criterion