API Reference / weka /

trees.py

Weka tree wrappers.

Decision and model trees from weka.classifiers.trees, registered under weka. hub keys so they sit alongside the native TuiML trees family without colliding with it.

Notes

Requires the optional Weka extra: pip install 'tuiml[weka]' plus a Java runtime (11+) on PATH.

Classes

J48

class weka.trees.J48(_WekaSupervisedMixin, Classifier)

J48 — Weka's C4.5 decision tree (hub key weka.J48).

Wraps weka.classifiers.trees.J48, Weka's implementation of Quinlan's C4.5. Splits on the attribute with the highest gain ratio and prunes with subtree raising by default.
Constructor 
__init__(
    self,
    confidence_factor: float = 0.25,
    min_num_obj: int = 2,
    unpruned: bool = False,
    binary_splits: bool = False,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

confidence_factor
float = 0.25
Confidence threshold for pruning (Weka -C). Lower values prune more aggressively.
min_num_obj
int = 2
Minimum number of instances per leaf (Weka -M).
unpruned
bool = False
Build an unpruned tree (Weka -U).
binary_splits
bool = False
Use binary splits on nominal attributes (Weka -B).
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical rather than numeric.
options
sequence of str or None = None
Extra raw Weka options appended to the generated ones, for anything not exposed as a named parameter.

Attributes

model_
weka.classifiers.Classifier
The fitted backing Weka classifier.
classes_
np.ndarray
Class labels in the caller's original label space.

See Also

DecisionTreeClassifier

TuiML's native decision tree (CART-style).
REPTree

Reduced-error pruning tree.
python 
>>> from tuiml.weka import J48
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = J48(confidence_factor=0.25).fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

REPTree

class weka.trees.REPTree(_WekaSupervisedMixin, Classifier)

REPTree — fast tree with reduced-error pruning (hub key weka.REPTree).

Wraps weka.classifiers.trees.REPTree. Builds a decision/regression tree using information gain (variance for numeric targets) and prunes it with reduced-error pruning and backfitting.
Constructor 
__init__(
    self,
    max_depth: int = ...,
    min_num: float = 2.0,
    num_folds: int = 3,
    no_pruning: bool = False,
    seed: int = 1,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

max_depth
int = -1
Maximum tree depth (Weka -L); -1 means unlimited.
min_num
float = 2.0
Minimum total instance weight in a leaf (Weka -M).
num_folds
int = 3
Folds used for pruning; one is held out (Weka -N).
no_pruning
bool = False
Disable pruning entirely (Weka -P).
seed
int = 1
Random seed (Weka -S).
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.

Attributes

model_
weka.classifiers.Classifier
The fitted backing Weka classifier.
python 
>>> from tuiml.weka import REPTree
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = REPTree(max_depth=5).fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

RandomTree

class weka.trees.RandomTree(_WekaSupervisedMixin, Classifier)

RandomTree — tree over a random attribute subset (hub key weka.RandomTree).

Wraps weka.classifiers.trees.RandomTree. Considers k randomly chosen attributes at each node and performs no pruning. Mainly used as the base learner inside RandomForest.
Constructor 
__init__(
    self,
    k_value: int = 0,
    max_depth: int = 0,
    min_num: float = 1.0,
    seed: int = 1,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

k_value
int = 0
Attributes sampled per split (Weka -K); 0 means :math:`\lfloor \log_2(p) \rfloor + 1`.
max_depth
int = 0
Maximum depth (Weka -depth); 0 means unlimited.
min_num
float = 1.0
Minimum total instance weight in a leaf (Weka -M).
seed
int = 1
Random seed (Weka -S).
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.
python 
>>> from tuiml.weka import RandomTree
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = RandomTree(seed=42).fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

RandomForest

class weka.trees.RandomForest(_WekaSupervisedMixin, Classifier)

RandomForest — Weka's random forest (hub key weka.RandomForest).

Wraps weka.classifiers.trees.RandomForest, a bagged ensemble of RandomTree.
Constructor 
__init__(
    self,
    num_iterations: int = 100,
    num_features: int = 0,
    max_depth: int = 0,
    seed: int = 1,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

num_iterations
int = 100
Number of trees (Weka -I).
num_features
int = 0
Attributes sampled per split (Weka -K); 0 means :math:`\lfloor \log_2(p) \rfloor + 1`, which is Weka's default and differs from scikit-learn's :math:`\sqrt{p}`.
max_depth
int = 0
Maximum depth (Weka -depth); 0 means unlimited.
seed
int = 1
Random seed (Weka -S).
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.

See Also

RandomForestClassifier

TuiML's native forest.
python 
>>> from tuiml.weka import RandomForest
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = RandomForest(num_iterations=20, seed=42).fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

DecisionStump

class weka.trees.DecisionStump(_WekaSupervisedMixin, Classifier)

DecisionStump — one-level decision tree (hub key weka.DecisionStump).

Wraps weka.classifiers.trees.DecisionStump. Splits on a single attribute; normally used as a weak learner inside a boosting ensemble.
Constructor 
__init__(
    self,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.
python 
>>> from tuiml.weka import DecisionStump
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = DecisionStump().fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

LMT

class weka.trees.LMT(_WekaSupervisedMixin, Classifier)

LMT — Logistic Model Tree (hub key weka.LMT).

Wraps weka.classifiers.trees.LMT. A decision tree carrying a logistic regression model at each leaf, fitted with LogitBoost.
Constructor 
__init__(
    self,
    min_num_instances: int = 15,
    fast_regression: bool = True,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

min_num_instances
int = 15
Minimum instances at which a node is considered for splitting (Weka -M).
fast_regression
bool = True
Use the heuristic that speeds up LogitBoost fitting. Passing False emits Weka's -R flag to disable it.
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.

Notes

LMT is considerably slower to train than J48, because every candidate node fits a boosted logistic model.
python 
>>> from tuiml.weka import LMT
>>> from tuiml.datasets import load_iris
>>> data = load_iris()
>>> clf = LMT().fit(data.X, data.y)
>>> clf.predict(data.X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.

M5P

class weka.trees.M5P(_WekaSupervisedMixin, Regressor)

M5P — M5 model tree for regression (hub key weka.M5P).

Wraps weka.classifiers.trees.M5P. Builds a regression tree whose leaves hold linear regression models rather than constants.
Constructor 
__init__(
    self,
    min_num_instances: float = 4.0,
    unpruned: bool = False,
    use_unsmoothed: bool = False,
    build_regression_tree: bool = False,
    nominal_features: Optional[Sequence[int]] = None,
    options: Optional[Sequence[str]] = None,
)

Parameters

min_num_instances
float = 4.0
Minimum instances allowed at a leaf (Weka -M).
unpruned
bool = False
Build an unpruned tree (Weka -N).
use_unsmoothed
bool = False
Skip the smoothing step applied to leaf predictions (Weka -U).
build_regression_tree
bool = False
Produce a plain regression tree — constants at the leaves instead of linear models (Weka -R).
nominal_features
sequence of int or None = None
0-based indices of columns to treat as categorical.
options
sequence of str or None = None
Extra raw Weka options.

See Also

DecisionTreeRegressor

TuiML's native regression tree (constant leaves, not linear models).
python 
>>> from tuiml.weka import M5P
>>> import numpy as np
>>> rng = np.random.default_rng(0)
>>> X = rng.normal(size=(80, 3))
>>> y = 2.0 * X[:, 0] + rng.normal(scale=0.1, size=80)
>>> reg = M5P().fit(X, y)
>>> reg.predict(X[:5]).shape
(5,)

Methods

get_parameter_schema (cls) -> Dict[str, Any]

Return JSON Schema for constructor parameters.

get_capabilities (cls) -> List[str]

Return the capability names this wrapper supports.