RuleTree.tree package

Contents

RuleTree.tree package#

Submodules#

RuleTree.tree.RuleTree module#

Base class of the RuleTree models: a decision tree built node by node, where each node is split by a stump chosen among the base stumps.

class RuleTree(max_leaf_nodes, min_samples_split, max_depth, prune_useless_leaves, base_stumps, stump_selection, random_state, exceptions='raise')[source]#

Bases: RuleTreeBase, ABC

Base class of the RuleTree models.

A RuleTree is a binary decision tree built best-first: the nodes are split one at a time, in the order given by queue_push, until a stopping criterion is met. Each split is a stump (see RuleTree.base.RuleTreeBaseStump.RuleTreeBaseStump) chosen among base_stumps, so the same tree can mix different kinds of splits (thresholds on features, distances from pivots, shapelets, …) and different data types (tabular data, time series, images).

The subclasses define the task: RuleTreeClassifier, RuleTreeRegressor, RuleTreeCluster and its variants.

Parameters:

  • max_leaf_nodes (int, default=None) – Maximum number of leaves. If None, unlimited.

  • min_samples_split (int) – Minimum number of instances to split a node.

  • max_depth (int) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list) –

    Stumps used to split the nodes. One of:

    • a stump;

    • a list of stumps, chosen with equal probability;

    • a list of tuples (probability, stump) or (probability, stump, data_types), where the probabilities sum to 1 and data_types (a data type or a list of data types, see RuleTree.utils.define.DATA_INPUT_TYPES) restricts the data the stump is used on. By default, a stump is used on every data type it supports. The probability of a stump is split equally among its data types.

  • stump_selection ({'random', 'best'}) – How the stump of each node is chosen: at random according to the probabilities, or the one with the highest gain among all the stumps and data types.

  • random_state (int, RandomState instance or None) – Seed of the random choice of the stumps, also given to the stumps whose random_state is None (so that they do not break ties at random).

  • exceptions ({'raise', 'ignore'}, default='raise') – What to do when a stump raises an error: raise it, or make the node a leaf and warn.

Variables:

  • root (RuleTreeNode) – Root of the fitted tree.

  • candidate_splits (list of tuple) – Pairs (stump, data type) usable on the training data, with their probability: (probability, stump, data_type).

  • cache (dict) – Values cached by the stumps during fit (see get_cache()).

See also

RuleTreeClassifier, RuleTreeRegressor, RuleTreeCluster

RuleTree.base.RuleTreeBaseStump.RuleTreeBaseStump

the interface of the stumps.

get_cache(namespace: str) → dict[source]#

Dictionary where the stumps store values shared among the splits, e.g. embeddings of the whole data set. The cache is cleared at every fit(), kept by partial_fit(), and can be cleared manually with clear_cache().

Parameters:

namespace (str) – Name of the cache. Each stump uses its cache_namespace(): its class name and a hash of its hyperparameters, e.g. get_cache(stump.cache_namespace()).

Returns:

cache – The cache of the namespace (created if missing).

Return type:

dict

clear_cache()[source]#

Remove all the values cached by the stumps (see get_cache()).

fit(X: ndarray | None = None, y: ndarray | None = None, X_ts: ndarray | None = None, X_img: ndarray | None = None, X_txt: ndarray | None = None, sample_weight=None)[source]#

Fit the tree on the provided data, starting from an empty tree.

At least one of X, X_ts, X_img and X_txt must be given; all the inputs refer to the same instances. Each argument is a data type (see RuleTree.utils.define.DATA_INPUT_TYPES): the stumps are used on the data types they support.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • y (numpy.ndarray of shape (n_samples,), default=None) – Target (None for clustering).

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

  • sample_weight (numpy.ndarray of shape (n_samples,), default=None) – Sample weights, passed to the stumps.

Returns:

self – The fitted tree.

Return type:

RuleTree

Raises:

ValueError – If no input is given, the inputs have different numbers of instances, or no stump supports the provided data.

partial_fit(X: ndarray | None = None, y: ndarray | None = None, X_ts: ndarray | None = None, X_img: ndarray | None = None, X_txt: ndarray | None = None, sample_weight=None)[source]#

Continue growing the fitted tree: the current leaves are split further using the provided data, within the limits of max_depth and max_leaf_nodes (e.g. after increasing them with set_params). The existing splits are not changed. If the tree is not fitted, this is fit().

The cache of the stumps (see get_cache()) is kept: call clear_cache() first if the data differ from those used to fill it.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • y (numpy.ndarray of shape (n_samples,), default=None) – Target (None for clustering).

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

  • sample_weight (numpy.ndarray of shape (n_samples,), default=None) – Sample weights, passed to the stumps.

Returns:

self – The fitted tree.

Return type:

RuleTree

make_split(data: dict, y: ndarray | None, idx: ndarray, sample_weight=None) → RuleTreeBaseStump[source]#

Fit the split of a node, choosing among candidate_splits_ according to stump_selection: a random pair (stump, data type), or the pair with the highest gain (gain_).

Parameters:

  • data (dict) – Whole data set, as {data_type: array}.

  • y (numpy.ndarray or None) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances in the node.

  • sample_weight (numpy.ndarray, default=None) – Sample weights of the whole data set.

Returns:

stump – The fitted stump.

Return type:

RuleTreeBaseStump

Raises:

NoSplitFoundWarning – If no stump can split the node.

is_split_useless(data: dict, idx: ndarray, labels: ndarray) → bool[source]#

Whether a split is useless, i.e. all the instances go to the same child.

Parameters:

  • data (dict) – Whole data set, as {data_type: array}.

  • idx (numpy.ndarray) – Indices of the instances in the node.

  • labels (numpy.ndarray) – Result of the split on idx (1 left, 2 right).

Returns:

useless

Return type:

bool

predict(X: ndarray = None, X_ts=None, X_img=None, X_txt=None)[source]#

Predict the output of each instance: the prediction of the leaf it reaches.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

Returns:

y_pred – Predicted classes, values or cluster labels.

Return type:

numpy.ndarray of shape (n_samples,)

apply(X: ndarray | None = None, X_ts=None, X_img=None, X_txt=None)[source]#

Leaf reached by each instance.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

Returns:

leaves – Identifier (node_id) of the leaf reached by each instance.

Return type:

numpy.ndarray of shape (n_samples,)

update_statistics(X: ndarray | None = None, y: ndarray | None = None, X_ts=None, X_img=None, X_txt=None)[source]#

Recompute the statistics of the nodes (predictions, probabilities, number of instances, …) and of their stumps on new data, without changing the splits. For classification, classes_ become the classes of y (e.g. to re-label the tree with a new target). A node reached by no instance predicts nan.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • y (numpy.ndarray of shape (n_samples,), default=None) – Target (None for clustering).

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

get_rules(columns_names=None)[source]#

Rules of the tree, as a nested dictionary.

Parameters:

columns_names (list of str, default=None) – Names of the tabular features, used in the rules instead of X_<index>.

Returns:

rules – Rule of the root: the node statistics (node_id, is_leaf, prediction, …), the rule of the split (textual_rule, not_textual_rule, …) and the rules of the children in left_node and right_node (None for a leaf). See print_rules().

Return type:

dict

make_leaf(node: RuleTreeNode) → RuleTreeNode[source]#

Hook called when a node becomes a leaf. The default implementation does nothing.

Parameters:

node (RuleTreeNode) – The new leaf.

Returns:

node

Return type:

RuleTreeNode

queue_pop()[source]#

Pop the next node to split from the queue.

Returns:

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node (RuleTreeNode)

check_additional_halting_condition(y, curr_idx: ndarray)[source]#

Additional condition to stop splitting a node, defined by the subclasses (e.g. a pure node for classification). The default implementation never stops.

Parameters:
Returns:

halt

Return type:

bool

abstractmethod prepare_node(y: ndarray | None, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) with the statistics of the instances idx: prediction, probabilities, number of instances.

Parameters:

  • y (numpy.ndarray or None) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node: its path from the root, e.g. "Rlr".

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

abstractmethod queue_push(node: RuleTreeNode, idx: ndarray)[source]#

Push a node to split into the queue: the priority defines the order in which the tree grows.

Parameters:
local_interpretation(X, joint_contribution=False)[source]#

Data needed to compute the contribution of each feature to the predictions (see the implementations of RuleTreeClassifier and RuleTreeRegressor, which return the contributions).

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features)) – Tabular data.

  • joint_contribution (bool, default=False) – Not used here.

Returns:

  • leaves (numpy.ndarray) – Index of the leaf reached by each instance.

  • paths (list of list of int) – Path from the root to each leaf.

  • leaf_to_path (dict) – Leaf index -> its path.

  • values (numpy.ndarray) – Values of the nodes (see _tree_value).

eval_contributions(leaves, paths, leaf_to_path, values, biases, line_shape, joint_contribution=False)[source]#

Contributions of the features to the predictions, as in treeinterpreter: the prediction of an instance is the value of the root (bias) plus the change of value along its path, attributed to the feature of each split.

Parameters:

  • leaves (numpy.ndarray) – Index of the leaf reached by each instance.

  • paths (list of list of int) – Path from the root to each leaf.

  • leaf_to_path (dict) – Leaf index -> its path.

  • values (numpy.ndarray) – Values of the nodes.

  • biases (numpy.ndarray) – Value of the root, for each instance.

  • line_shape (int or tuple) – Shape of the contributions of one instance.

  • joint_contribution (bool, default=False) – If True, the contributions are given to the sets of features along the path.

Returns:

  • direct_prediction (numpy.ndarray) – Value of the leaf of each instance.

  • biases (numpy.ndarray) – Value of the root.

  • contributions (numpy.ndarray or list of dict) – Contribution of each feature (or set of features, with joint_contribution) to each prediction.

classmethod print_rules(rules: dict = None, digits=2, indent: int = 0)[source]#

Print rules as an indented list of conditions and outputs.

Parameters:

  • rules (dict) – Rules returned by get_rules().

  • digits (int, default=2) – Number of decimals of the numerical outputs.

  • indent (int, default=0) – Indentation level of the root.

to_dict(filename=None)[source]#

Dictionary representation of the fitted tree, from which from_dict() rebuilds it.

Parameters:

filename (str or path-like, default=None) – If given, the dictionary is also saved to this file as JSON.

Returns:

dictionary – JSON-friendly dictionary with the class of the tree, its parameters (args, base stumps included), the classes for classification trees and the nodes. The values may be NumPy scalars: to write it with json, use cls=RuleTree.utils.data_utils.json_numpy_encoder (as done with filename). See Saving and loading trees.

Return type:

dict

Raises:

sklearn.exceptions.NotFittedError – If the tree is not fitted.

classmethod from_dict(filename)[source]#

Rebuild a tree from a dictionary created by to_dict().

Parameters:

filename (dict, str or path-like) – The dictionary, or the JSON file where it is saved.

Returns:

tree – The tree, of the class it had when saved.

Return type:

RuleTree

export_graphviz(columns_names=None, scaler=None, float_precision=3, filename: str | None = None)[source]#

Graphviz representation of the tree (requires the graphviz package).

Parameters:

  • columns_names (list of str, default=None) – Names of the tabular features.

  • scaler (TransformerMixin, default=None) – Scaler used to preprocess the data: the thresholds are shown in the original scale.

  • float_precision (int, default=3) – Number of decimals of the thresholds.

  • filename (str, default=None) – If given, the graph is rendered to this file; otherwise it is rendered to a temporary file and opened (both require the Graphviz binaries).

Returns:

graph – The graph, if filename is None.

Return type:

graphviz.Digraph or None

get_predicates()[source]#

Internal nodes (the ones with a split) in depth-first order.

Returns:

predicates – Node id -> RuleTreeNode.

Return type:

dict

get_node_by_id(node_id)[source]#

Node with the given identifier.

Parameters:

node_id (str) – Identifier of the node: its path from the root, e.g. "Rlr".

Returns:

node – The node, or None if not in the tree.

Return type:

RuleTreeNode or None

get_leaf_nodes()[source]#

Leaves of the tree.

Returns:

leaves – Node id -> RuleTreeNode.

Return type:

dict

set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTree#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTree#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTree#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

RuleTree.tree.RuleTreeClassifier module#

RuleTreeClassifier: the RuleTree for classification.

class RuleTreeClassifier(max_leaf_nodes=inf, min_samples_split=2, max_depth=inf, prune_useless_leaves=False, base_stumps: ClassifierMixin | list = None, stump_selection: str = 'random', random_state=None)[source]#

Bases: ClassifierMixin, RuleTree

RuleTree for classification.

The tree grows breadth-first; a node is not split if it is pure. Each node predicts the majority class of its instances and stores the class probabilities and log-odds.

Parameters:

  • max_leaf_nodes (int, default=inf) – Maximum number of leaves.

  • min_samples_split (int, default=2) – Minimum number of instances to split a node.

  • max_depth (int, default=inf) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool, default=False) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list, default=None) – Stumps used to split the nodes (see RuleTree.tree.RuleTree.RuleTree); they must be classification stumps. If None, DecisionTreeStumpClassifier().

  • stump_selection ({'random', 'best'}, default='random') – How the stump of each node is chosen among base_stumps.

  • random_state (int, RandomState instance or None, default=None) – Seed of the random choice of the stumps, also given to the stumps without a random_state.

Variables:

  • classes (numpy.ndarray) – Classes of the target.

  • n_classes (int) – Number of classes.

  • root (RuleTreeNode) – Root of the fitted tree.

  • candidate_splits (list of tuple) – (probability, stump, data_type): the stumps usable on the training data, with the data type they are used on.

  • cache (dict) – Values cached by the stumps during fit (see get_cache()).

See also

RuleTreeRegressor

the same tree for regression.

RuleTree.stumps.classification

the stumps for classification.

Examples

>>> from sklearn.datasets import load_iris
>>> from RuleTree import RuleTreeClassifier
>>> X, y = load_iris(return_X_y=True)
>>> model = RuleTreeClassifier(max_depth=2, random_state=0).fit(X, y)
>>> model.predict(X[:3])
array([0, 0, 0])
check_additional_halting_condition(y, curr_idx: ndarray)[source]#

Stop splitting a node whose instances all have the same class.

Parameters:
Returns:

halt – True if the node is pure.

Return type:

bool

queue_push(node: RuleTreeNode, idx: ndarray)[source]#

Push a node into the queue, with priority to the shallowest nodes (breadth-first growth).

Parameters:
prepare_node(y: ndarray, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) with the majority class, the class probabilities, the log-odds and the number of the instances idx.

Parameters:

  • y (numpy.ndarray) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node.

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

predict_proba(X: ndarray = None, X_ts=None, X_img=None, X_txt=None)[source]#

Class probabilities of each instance: the ones of the leaf it reaches.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • X_ts (numpy.ndarray of shape (n_samples, n_signals, length), default=None) – Time series.

  • X_img (numpy.ndarray of shape (n_samples, ...), default=None) – Images.

  • X_txt (array-like of shape (n_samples,), default=None) – Texts.

Returns:

proba – Probabilities of the classes, in the order of classes_.

Return type:

numpy.ndarray of shape (n_samples, n_classes)

local_interpretation(X, joint_contribution=False)[source]#

Contribution of each feature to the predicted class probabilities, as in treeinterpreter: the probabilities of an instance are the ones of the root (bias) plus the contributions of the features of the splits along its path.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features)) – Tabular data.

  • joint_contribution (bool, default=False) – If True, the contributions are given to the sets of features along the path.

Returns:

  • prediction (numpy.ndarray of shape (n_samples, n_classes)) – Class probabilities of the leaf of each instance.

  • bias (numpy.ndarray of shape (n_samples, n_classes)) – Class probabilities of the root.

  • contributions (numpy.ndarray of shape (n_samples, n_features, n_classes) or list of dict) – Contribution of each feature (or set of features, with joint_contribution).

get_balanced_stumps(current_node=None, stumps=None, p=0.2)[source]#

Internal nodes whose split is balanced: the smaller side has a fraction of the training set (balance_score) above p. The visit stops at the first unbalanced node of each path.

Parameters:

  • current_node (RuleTreeNode, default=None) – Root of the subtree. If None, the root of the tree.

  • stumps (dict, default=None) – Dictionary being filled.

  • p (float, default=0.2) – Minimum balance score.

Returns:

stumps – Node id -> (node, balance score).

Return type:

dict

stumps_to_trees(balanced_nodes)[source]#

A tree of depth 1 for each of the given nodes: its split and two leaves.

Parameters:

balanced_nodes (dict) – Node id -> (node, balance score), as returned by get_balanced_stumps().

Returns:

trees – (feature, threshold) of the split -> RuleTreeClassifier with that split as root.

Return type:

dict

classmethod complete_tree(node, X, y, n_classes_)[source]#

Recompute recursively the predictions, class probabilities and number of instances of the nodes of a tree on new data, with classes 0, ..., n_classes_ - 1.

Parameters:
set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClassifier#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClassifier#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_proba_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeClassifier#

Configure whether metadata should be requested to be passed to the predict_proba method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict_proba if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict_proba.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict_proba.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict_proba.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict_proba.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeClassifier#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClassifier#

Configure whether metadata should be requested to be passed to the score method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to score.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in score.

Returns:

self – The updated object.

Return type:

object

RuleTree.tree.RuleTreeCluster module#

RuleTreeCluster and its variants: RuleTrees for clustering, also used to classify or regress with clusters found without looking at the target.

class RuleTreeCluster(n_components: int = 2, clus_impurity: str = 'r2', bic_eps: float = 0.0, max_leaf_nodes=inf, min_samples_split=2, max_depth=inf, prune_useless_leaves=False, base_stumps: RegressorMixin | list = None, random_state=None)[source]#

Bases: ClusterMixin, RuleTree

RuleTree for clustering: each leaf is a cluster described by the rules of its path.

To split a node, the instances are projected on their principal components (PCA, MCA or FAMD, for numerical, categorical or mixed data); for each component, a regression stump is fitted to predict it, and the best split is kept according to clus_impurity. A node is split only if the split decreases its BIC (see bic_eps). The tree grows from the largest node. Only tabular data are supported.

After fit, the nodes predict integer labels assigned in breadth-first order (‘R’ -> 0, ‘Rl’ -> 1, ‘Rr’ -> 2, …); predict returns the label of the leaf of each instance.

Parameters:

  • n_components (int, default=2) – Number of principal components predicted by the stumps of each node.

  • clus_impurity ({'r2', 'bic'}, default='r2') – Criterion to choose the split among the ones on the different components: the variance of the component explained by the split (‘r2’) or the Bayesian Information Criterion of the two clusters (‘bic’).

  • bic_eps (float, default=0.0) – A split is kept only if it decreases the BIC of the node by more than bic_eps times its value.

  • max_leaf_nodes (int, default=inf) – Maximum number of leaves.

  • min_samples_split (int, default=2) – Minimum number of instances to split a node.

  • max_depth (int, default=inf) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool, default=False) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list, default=None) – Regression stumps used to split the nodes, fitted to predict a principal component. If None, DecisionTreeStumpRegressor().

  • random_state (int, RandomState instance or None, default=None) – Seed of the principal component analysis and of the random choice of the stumps, also given to the stumps without a random_state.

Variables:

  • label_encoder (dict) – Node id -> integer label.

  • root (RuleTreeNode) – Root of the fitted tree.

  • candidate_splits (list of tuple) – (probability, stump, data_type): the stumps usable on the training data, with the data type they are used on.

  • cache (dict) – Values cached by the stumps during fit (see get_cache()).

See also

RuleTreeClusterClassifier, RuleTreeClusterRegressor

Examples

>>> from sklearn.datasets import make_blobs
>>> from sklearn.metrics import adjusted_rand_score
>>> from RuleTree import RuleTreeCluster
>>> X, y = make_blobs(n_samples=300, centers=3, cluster_std=0.5, random_state=0)
>>> labels = RuleTreeCluster(max_depth=3, random_state=0).fit_predict(X)
>>> round(adjusted_rand_score(y, labels), 2)
0.95
is_split_useless(data: dict, idx: ndarray, labels: ndarray)[source]#

Whether a split is useless: all the instances go to the same child, or the BIC of the two clusters is not lower than the BIC of the node (by bic_eps times its value).

Parameters:

  • data (dict) – Whole data set, as {data_type: array}.

  • idx (numpy.ndarray) – Indices of the instances in the node.

  • labels (numpy.ndarray) – Result of the split on idx (1 left, 2 right).

Returns:

useless

Return type:

bool

fit_predict(X, y=None, **kwargs)[source]#

Fit the tree and return the cluster label of each instance.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features)) – Tabular data.

  • y (None) – Not used, present for API consistency.

  • **kwargs (dict) – Arguments passed to fit.

Returns:

labels – Cluster labels.

Return type:

numpy.ndarray of shape (n_samples,)

queue_push(node: RuleTreeNode, idx: ndarray)[source]#

Push a node into the queue, with priority to the nodes with more instances.

Parameters:
make_split(data: dict, y, idx: ndarray, sample_weight=None) → RuleTreeBaseStump[source]#

Split a node on its principal components: the instances are projected with PCA, MCA or FAMD (numerical, categorical or mixed data), a stump is fitted to predict each component, and the best split according to clus_impurity is kept.

Parameters:

  • data (dict) – Whole data set, as {data_type: array} (only tabular data).

  • y (numpy.ndarray or None) – Not used by the split (the stumps predict the components).

  • idx (numpy.ndarray) – Indices of the instances in the node.

  • sample_weight (None) – Not used, present for API consistency.

Returns:

stump – The best stump.

Return type:

RuleTreeBaseStump

Raises:

NoSplitFoundWarning – If no component can be split.

prepare_node(y: ndarray, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) whose prediction is its node id, replaced by an integer label in _post_fit_fix().

Parameters:

  • y (numpy.ndarray or None) – Not used.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node.

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeCluster#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeCluster#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeCluster#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

class RuleTreeClusterClassifier(n_components: int = 2, clus_impurity: str = 'r2', bic_eps: float = 0.0, max_leaf_nodes=inf, min_samples_split=2, max_depth=inf, prune_useless_leaves=False, base_stumps: RegressorMixin | list = None, random_state=None)[source]#

Bases: ClassifierMixin, RuleTreeCluster

RuleTreeCluster whose leaves predict a class: the tree is built as in RuleTreeCluster, without looking at the target, and each node predicts the majority class of its instances, as in RuleTreeClassifier.

Parameters:

  • n_components (int, default=2) – Number of principal components predicted by the stumps of each node.

  • clus_impurity ({'r2', 'bic'}, default='r2') – Criterion to choose the split among the ones on the different components: the variance of the component explained by the split (‘r2’) or the Bayesian Information Criterion of the two clusters (‘bic’).

  • bic_eps (float, default=0.0) – A split is kept only if it decreases the BIC of the node by more than bic_eps times its value.

  • max_leaf_nodes (int, default=inf) – Maximum number of leaves.

  • min_samples_split (int, default=2) – Minimum number of instances to split a node.

  • max_depth (int, default=inf) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool, default=False) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list, default=None) – Regression stumps used to split the nodes, fitted to predict a principal component. If None, DecisionTreeStumpRegressor().

  • random_state (int, RandomState instance or None, default=None) – Seed of the principal component analysis and of the random choice of the stumps, also given to the stumps without a random_state.

Variables:

  • classes (numpy.ndarray) – Classes of the target.

  • n_classes (int) – Number of classes.

See also

RuleTreeCluster

the clustering tree.

RuleTreeClassifier

a tree whose splits are chosen with the target.

Examples

>>> from sklearn.datasets import make_blobs
>>> from RuleTree.tree.RuleTreeCluster import RuleTreeClusterClassifier
>>> X, y = make_blobs(n_samples=300, centers=3, cluster_std=0.5, random_state=0)
>>> model = RuleTreeClusterClassifier(max_depth=2, random_state=0).fit(X, y)
>>> round(model.score(X, y), 2)
0.99
prepare_node(y: ndarray, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) with the statistics of classification, as in RuleTreeClassifier.

Parameters:

  • y (numpy.ndarray) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node.

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

predict_proba(X: ndarray = None, X_ts=None, X_img=None, X_txt=None)[source]#

Class probabilities of each instance: the ones of the leaf it reaches.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features), default=None) – Tabular data.

  • X_ts (default=None) – Not supported by clustering, present for API consistency.

  • X_img (default=None) – Not supported by clustering, present for API consistency.

  • X_txt (default=None) – Not supported by clustering, present for API consistency.

Returns:

proba

Return type:

numpy.ndarray of shape (n_samples, n_classes)

set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterClassifier#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterClassifier#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_proba_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeClusterClassifier#

Configure whether metadata should be requested to be passed to the predict_proba method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict_proba if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict_proba.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict_proba.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict_proba.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict_proba.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeClusterClassifier#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterClassifier#

Configure whether metadata should be requested to be passed to the score method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to score.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in score.

Returns:

self – The updated object.

Return type:

object

class RuleTreeClusterRegressor(n_components: int = 2, clus_impurity: str = 'r2', bic_eps: float = 0.0, max_leaf_nodes=inf, min_samples_split=2, max_depth=inf, prune_useless_leaves=False, base_stumps: RegressorMixin | list = None, random_state=None)[source]#

Bases: RegressorMixin, RuleTreeCluster

RuleTreeCluster whose leaves predict a value: the tree is built as in RuleTreeCluster, without looking at the target, and each node predicts the mean of the target of its instances, as in RuleTreeRegressor.

Parameters:

  • n_components (int, default=2) – Number of principal components predicted by the stumps of each node.

  • clus_impurity ({'r2', 'bic'}, default='r2') – Criterion to choose the split among the ones on the different components: the variance of the component explained by the split (‘r2’) or the Bayesian Information Criterion of the two clusters (‘bic’).

  • bic_eps (float, default=0.0) – A split is kept only if it decreases the BIC of the node by more than bic_eps times its value.

  • max_leaf_nodes (int, default=inf) – Maximum number of leaves.

  • min_samples_split (int, default=2) – Minimum number of instances to split a node.

  • max_depth (int, default=inf) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool, default=False) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list, default=None) – Regression stumps used to split the nodes, fitted to predict a principal component. If None, DecisionTreeStumpRegressor().

  • random_state (int, RandomState instance or None, default=None) – Seed of the principal component analysis and of the random choice of the stumps, also given to the stumps without a random_state.

Variables:

root (RuleTreeNode) – Root of the fitted tree.

See also

RuleTreeCluster

the clustering tree.

RuleTreeRegressor

a tree whose splits are chosen with the target.

Examples

>>> from sklearn.datasets import load_diabetes
>>> from RuleTree.tree.RuleTreeCluster import RuleTreeClusterRegressor
>>> X, y = load_diabetes(return_X_y=True)
>>> model = RuleTreeClusterRegressor(max_depth=2, random_state=0).fit(X, y)
>>> model.predict(X[:2]).round(1)
array([177.6, 111.8])
prepare_node(y: ndarray, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) with the statistics of regression, as in RuleTreeRegressor.

Parameters:

  • y (numpy.ndarray) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node.

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterRegressor#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterRegressor#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeClusterRegressor#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeClusterRegressor#

Configure whether metadata should be requested to be passed to the score method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to score.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in score.

Returns:

self – The updated object.

Return type:

object

RuleTree.tree.RuleTreeNode module#

RuleTreeNode: a node of a RuleTree.

class RuleTreeNode(node_id: str, prediction: int | str | float, prediction_probability: ndarray | float, log_odds: ndarray | float, classes: ndarray, parent: RuleTreeNode | None, stump: RuleTreeBaseStump = None, node_l: RuleTreeNode = None, node_r: RuleTreeNode = None, **kwargs)[source]#

Bases: object

A node of a RuleTree: a leaf with a prediction, or an internal node with a stump that splits its instances between its two children.

The identifier of a node is its path from the root: "R" is the root, "Rl" and "Rr" its left and right children, "Rlr" the right child of "Rl", and so on.

Parameters:

  • node_id (str) – Identifier of the node.

  • prediction (int, str or float) – Predicted class, value or cluster label.

  • prediction_probability (numpy.ndarray or float) – Class probabilities (classification) or standard deviation of the target (regression).

  • log_odds (numpy.ndarray or float) – Log-odds of the classes (classification).

  • classes (numpy.ndarray or None) – Classes of the target (classification), None otherwise.

  • parent (RuleTreeNode or None) – Parent node (None for the root).

  • stump (RuleTreeBaseStump, default=None) – Split of the node (None for a leaf).

  • node_l (RuleTreeNode, default=None) – Left and right children (None for a leaf).

  • node_r (RuleTreeNode, default=None) – Left and right children (None for a leaf).

  • **kwargs (dict) – Other attributes, e.g. samples (number of instances).

Variables:

balance_score (float) – For internal nodes, fraction of the training set in the smaller side of the split.

Examples

>>> from sklearn.datasets import load_iris
>>> from RuleTree import RuleTreeClassifier
>>> X, y = load_iris(return_X_y=True)
>>> model = RuleTreeClassifier(max_depth=2, random_state=0).fit(X, y)
>>> node = model.get_node_by_id("Rr")
>>> node.get_depth(), node.samples, node.is_leaf()
(1, 100, False)
>>> node.stump.get_rule()["textual_rule"], node.node_l.node_id
('X_3 <= 1.75', 'Rrl')
is_leaf()[source]#

Whether the node is a leaf (it has no children).

Returns:

is_leaf

Return type:

bool

make_leaf()[source]#

Make the node a leaf by removing its children.

Returns:

self

Return type:

RuleTreeNode

simplify() → RuleTreeNode[source]#

Merge the subtrees whose leaves all have the same prediction (see prune_useless_leaves).

Returns:

self – The root of the simplified subtree.

Return type:

RuleTreeNode

set_stump(stump: RuleTreeBaseStump)[source]#

Set the split of the node.

Parameters:

stump (RuleTreeBaseStump) – The fitted stump.

get_possible_outputs() → tuple[set, set][source]#

Predictions of the nodes of the subtree.

Returns:

  • leaf_predictions (set) – Predictions of the leaves.

  • internal_predictions (set) – Predictions of the internal nodes.

get_depth()[source]#

Depth of the node, from its identifier (the root has depth 0).

Returns:

depth

Return type:

int

get_rule(columns_names: list = None, scaler: TransformerMixin = None)[source]#

Rules of the subtree, as a nested dictionary (see RuleTree.get_rules).

Parameters:

  • columns_names (list of str, default=None) – Names of the tabular features.

  • scaler (TransformerMixin, default=None) – Scaler used to preprocess the data: the thresholds are shown in the original scale.

Returns:

rule – Statistics of the node (node_id, is_leaf, prediction, prediction_probability, log_odds, prediction_classes_), the rules of the children (left_node, right_node) and, for an internal node, the rule of its stump.

Return type:

dict

update_statistics(data, y, idx=None, context=None)[source]#

Recompute the statistics of classification of the node (majority class, class probabilities, log-odds, number of instances) on the instances idx, and the statistics of its stump. A node without instances predicts nan.

Parameters:

  • data (dict) – Whole data set, as {data_type: array}.

  • y (numpy.ndarray) – Target of the whole data set.

  • idx (numpy.ndarray, default=None) – Indices of the instances reaching the node. If None, all the instances.

  • context (RuleTree, default=None) – The tree, whose classes_ are used when the node has none.

node_to_dict()[source]#

Dictionary representation of the node (without its children, referenced by id).

Returns:

node_dict – JSON-friendly dictionary with the statistics of the node (also samples and, for an internal node, balance_score), the ids of its children and, for an internal node, the dictionary of its stump.

Return type:

dict

classmethod dict_to_node(info_dict)[source]#

Create a node from a dictionary created by node_to_dict() (children not linked).

Parameters:

info_dict (dict) – Dictionary representation of the node.

Returns:

node

Return type:

RuleTreeNode

encode_node(index, parent, vector, stump, node_index=0)[source]#

Encode the subtree in arrays indexed as a complete binary tree (children of i at 2i+1 and 2i+2), for trees of univariate splits.

Parameters:

  • index (dict) – Node id -> position, filled.

  • parent (numpy.ndarray) – Position of the parent of each position, filled.

  • vector (numpy.ndarray of shape (2, n_positions)) – Feature + 1 (-1 for leaves) and threshold (prediction for leaves) of each position, filled.

  • stump (object) – Not used.

  • node_index (int, default=0) – Position of this node.

export_graphviz(graph=None, columns_names=None, scaler=None, float_precision=3)[source]#

Graphviz representation of the subtree (requires the graphviz package).

Parameters:

  • graph (graphviz.Digraph, default=None) – Graph to add the nodes to. If None, a new graph.

  • columns_names (list of str, default=None) – Names of the tabular features.

  • scaler (TransformerMixin, default=None) – Scaler used to preprocess the data: the thresholds are shown in the original scale.

  • float_precision (int, default=3) – Number of decimals of the thresholds.

Returns:

graph

Return type:

graphviz.Digraph

predict(data, idx: ndarray = None, prediction_step=inf)[source]#

Predictions, leaves and probabilities of the instances, traversing the subtree.

Parameters:

  • data (dict or numpy.ndarray) – Data set, as {data_type: array}. A single array is passed as is to the stumps.

  • idx (numpy.ndarray, default=None) – Indices of the instances to predict. If None, all the instances.

  • prediction_step (int, default=numpy.inf) – Maximum number of splits to traverse: with a finite value, the prediction of the node reached after that many splits.

Returns:

  • labels (numpy.ndarray of shape (n_samples,)) – Prediction of each instance.

  • leaves (numpy.ndarray of shape (n_samples,)) – Id of the node reached by each instance.

  • proba (numpy.ndarray of shape (n_samples, n_values)) – prediction_probability of the node reached by each instance.

get_predicates() → dict[source]#

Internal nodes of the subtree in depth-first order.

Returns:

predicates – Node id -> RuleTreeNode.

Return type:

dict

get_node_by_id(node_id: str) → RuleTreeNode | None[source]#

Node of the subtree with the given identifier.

Parameters:

node_id (str) – Identifier of the node.

Returns:

node – The node, or None if not in the subtree.

Return type:

RuleTreeNode or None

get_leaf_nodes() → dict[source]#

Leaves of the subtree.

Returns:

leaves – Node id -> RuleTreeNode.

Return type:

dict

stump_class(stump_dict: dict) → type[source]#

Class of a stump saved by node_to_dict: stump_class in module stump_type.

Parameters:

stump_dict (dict) – Dictionary representation of the stump.

Returns:

stump_class

Return type:

type

RuleTree.tree.RuleTreeRegressor module#

RuleTreeRegressor: the RuleTree for regression.

class RuleTreeRegressor(max_leaf_nodes=inf, min_samples_split=2, max_depth=inf, prune_useless_leaves=False, base_stumps: RegressorMixin | list = None, stump_selection: str = 'random', random_state=None)[source]#

Bases: RegressorMixin, RuleTree

RuleTree for regression.

The tree grows breadth-first. Each node predicts the mean of the target of its instances and stores its standard deviation in prediction_probability.

Parameters:

  • max_leaf_nodes (int, default=inf) – Maximum number of leaves.

  • min_samples_split (int, default=2) – Minimum number of instances to split a node.

  • max_depth (int, default=inf) – Maximum depth of the tree (the root has depth 0).

  • prune_useless_leaves (bool, default=False) – Whether to merge, after fit, the sibling leaves with the same prediction.

  • base_stumps (RuleTreeBaseStump or list, default=None) – Stumps used to split the nodes (see RuleTree.tree.RuleTree.RuleTree); they must be regression stumps. If None, DecisionTreeStumpRegressor().

  • stump_selection ({'random', 'best'}, default='random') – How the stump of each node is chosen among base_stumps.

  • random_state (int, RandomState instance or None, default=None) – Seed of the random choice of the stumps, also given to the stumps without a random_state.

Variables:

  • root (RuleTreeNode) – Root of the fitted tree.

  • candidate_splits (list of tuple) – (probability, stump, data_type): the stumps usable on the training data, with the data type they are used on.

  • cache (dict) – Values cached by the stumps during fit (see get_cache()).

See also

RuleTreeClassifier

the same tree for classification.

RuleTree.stumps.regression

the stumps for regression.

Examples

>>> from sklearn.datasets import load_diabetes
>>> from RuleTree import RuleTreeRegressor
>>> X, y = load_diabetes(return_X_y=True)
>>> model = RuleTreeRegressor(max_depth=3, random_state=0).fit(X, y)
>>> model.predict(X[:2]).round(1)
array([208.6,  83.4])
queue_push(node: RuleTreeNode, idx: ndarray)[source]#

Push a node into the queue, with priority to the shallowest nodes (breadth-first growth).

Parameters:
prepare_node(y: ndarray, idx: ndarray, node_id: str, node: RuleTreeNode | None = None) → RuleTreeNode[source]#

Create a node (or update node) with the mean (prediction) and the standard deviation of the target and the number of the instances idx.

Parameters:

  • y (numpy.ndarray) – Target of the whole data set.

  • idx (numpy.ndarray) – Indices of the instances reaching the node.

  • node_id (str) – Identifier of the node.

  • node (RuleTreeNode, default=None) – Existing node to update instead of creating a new one.

Returns:

node

Return type:

RuleTreeNode

local_interpretation(X, joint_contribution=False)[source]#

Contribution of each feature to the predictions, as in treeinterpreter: the prediction of an instance is the one of the root (bias) plus the contributions of the features of the splits along its path.

Parameters:

  • X (numpy.ndarray of shape (n_samples, n_features)) – Tabular data.

  • joint_contribution (bool, default=False) – If True, the contributions are given to the sets of features along the path.

Returns:

  • prediction (numpy.ndarray of shape (n_samples,)) – Prediction of the leaf of each instance.

  • bias (numpy.ndarray of shape (n_samples,)) – Prediction of the root.

  • contributions (numpy.ndarray of shape (n_samples, n_features) or list of dict) – Contribution of each feature (or set of features, with joint_contribution).

set_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeRegressor#

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_partial_fit_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$', sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeRegressor#

Configure whether metadata should be requested to be passed to the partial_fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to partial_fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to partial_fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in partial_fit.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in partial_fit.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in partial_fit.

  • sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in partial_fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, X_img: bool | None | str = '$UNCHANGED$', X_ts: bool | None | str = '$UNCHANGED$', X_txt: bool | None | str = '$UNCHANGED$') → RuleTreeRegressor#

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

  • X_img (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_img parameter in predict.

  • X_ts (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_ts parameter in predict.

  • X_txt (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for X_txt parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, sample_weight: bool | None | str = '$UNCHANGED$') → RuleTreeRegressor#

Configure whether metadata should be requested to be passed to the score method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config()). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to score.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:

sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in score.

Returns:

self – The updated object.

Return type:

object

Module contents#

Contents