sklearn: Add HDBSCAN, Tree node-array properties, and precision_recall_curve drop_intermediate/y_score

stubs/sklearn/cluster/__init__.pyi

@@ -9,6 +9,7 @@ from ._bicluster import SpectralBiclustering as SpectralBiclustering, SpectralCo
 from ._birch import Birch as Birch
 from ._bisect_k_means import BisectingKMeans as BisectingKMeans
 from ._dbscan import DBSCAN as DBSCAN, dbscan as dbscan
+from ._hdbscan.hdbscan import HDBSCAN as HDBSCAN
 from ._kmeans import KMeans as KMeans, MiniBatchKMeans as MiniBatchKMeans, k_means as k_means, kmeans_plusplus as kmeans_plusplus
 from ._mean_shift import (
     MeanShift as MeanShift,
@@ -29,6 +30,7 @@ __all__ = [
     "AgglomerativeClustering",
     "Birch",
     "DBSCAN",
+    "HDBSCAN",
     "OPTICS",
     "cluster_optics_dbscan",
     "cluster_optics_xi",

stubs/sklearn/cluster/_hdbscan/hdbscan.pyi

@@ -0,0 +1,38 @@
+from typing import Any, Callable, ClassVar, Literal
+from typing_extensions import Self
+
+from numpy import ndarray
+
+from ..._typing import Float, Int, MatrixLike
+from ...base import BaseEstimator, ClusterMixin
+
+class HDBSCAN(ClusterMixin, BaseEstimator):
+    labels_: ndarray = ...
+    probabilities_: ndarray = ...
+    n_features_in_: int = ...
+    feature_names_in_: ndarray = ...
+    centroids_: ndarray = ...
+    medoids_: ndarray = ...
+
+    _parameter_constraints: ClassVar[dict] = ...
+
+    def __init__(
+        self,
+        min_cluster_size: Int = 5,
+        min_samples: None | Int = None,
+        cluster_selection_epsilon: Float = 0.0,
+        max_cluster_size: None | Int = None,
+        metric: str | Callable = "euclidean",
+        metric_params: None | dict = None,
+        alpha: Float = 1.0,
+        algorithm: Literal["auto", "brute", "kd_tree", "ball_tree"] = "auto",
+        leaf_size: Int = 40,
+        n_jobs: None | Int = None,
+        cluster_selection_method: Literal["eom", "leaf"] = "eom",
+        allow_single_cluster: bool = False,
+        store_centers: None | Literal["centroid", "medoid", "both"] = None,
+        copy: bool = False,
+    ) -> None: ...
+    def fit(self, X: MatrixLike, y: Any = None) -> Self: ...
+    def fit_predict(self, X: MatrixLike, y: Any = None) -> ndarray: ...
+    def dbscan_clustering(self, cut_distance: Float, min_cluster_size: Int = 5) -> ndarray: ...

stubs/sklearn/metrics/_ranking.pyi

@@ -1,4 +1,5 @@
-from typing import Literal
+from typing import Literal, overload
+from typing_extensions import deprecated
 
 from numpy import ndarray
 
@@ -40,12 +41,25 @@ def roc_auc_score(
     multi_class: Literal["raise", "ovr", "ovo"] = "raise",
     labels: None | ArrayLike = None,
 ) -> Float: ...
+@overload
 def precision_recall_curve(
     y_true: ArrayLike,
-    probas_pred: ArrayLike,
+    y_score: ArrayLike,
     *,
     pos_label: None | str | Int = None,
     sample_weight: None | ArrayLike = None,
+    drop_intermediate: bool = False,
+) -> tuple[ndarray, ndarray, ndarray]: ...
+@deprecated("`probas_pred` is deprecated in 1.5 and will be removed in 1.7. Use `y_score` instead.")
+@overload
+def precision_recall_curve(
+    y_true: ArrayLike,
+    y_score: None = None,
+    *,
+    pos_label: None | str | Int = None,
+    sample_weight: None | ArrayLike = None,
+    drop_intermediate: bool = False,
+    probas_pred: ArrayLike,
 ) -> tuple[ndarray, ndarray, ndarray]: ...
 def roc_curve(
     y_true: ArrayLike,

stubs/sklearn/stubtest_allowlist.txt

@@ -95,14 +95,13 @@ sklearn.cluster.BisectingKMeans.set_fit_request
 sklearn.cluster.BisectingKMeans.set_score_request
 sklearn.cluster.DBSCAN.__sklearn_tags__
 sklearn.cluster.DBSCAN.set_fit_request
-sklearn.cluster.HDBSCAN
+sklearn.cluster.HDBSCAN.__sklearn_tags__
 sklearn.cluster.KMeans.set_fit_request
 sklearn.cluster.KMeans.set_score_request
 sklearn.cluster.MiniBatchKMeans.set_fit_request
 sklearn.cluster.MiniBatchKMeans.set_partial_fit_request
 sklearn.cluster.MiniBatchKMeans.set_score_request
 sklearn.cluster.SpectralClustering.__sklearn_tags__
-sklearn.cluster.__all__
 sklearn.cluster._affinity_propagation.AffinityPropagation.__sklearn_tags__
 sklearn.cluster._bicluster.BaseSpectral.__sklearn_tags__
 sklearn.cluster._birch.Birch.__sklearn_tags__
@@ -112,7 +111,9 @@ sklearn.cluster._bisect_k_means.BisectingKMeans.set_score_request
 sklearn.cluster._dbscan.DBSCAN.__sklearn_tags__
 sklearn.cluster._dbscan.DBSCAN.set_fit_request
 sklearn.cluster._dbscan_inner.__test__
-sklearn.cluster._hdbscan.hdbscan
+sklearn.cluster._hdbscan.hdbscan.FAST_METRICS
+sklearn.cluster._hdbscan.hdbscan.HDBSCAN.__sklearn_tags__
+sklearn.cluster._hdbscan.hdbscan.remap_single_linkage_tree
 sklearn.cluster._hdbscan.tests
 sklearn.cluster._hdbscan.tests.test_reachibility
 sklearn.cluster._k_means_common.__pyx_capi__
@@ -1229,17 +1230,7 @@ sklearn.tree._splitter.RandomSplitter
 sklearn.tree._splitter.__test__
 sklearn.tree._tree.NODE_DTYPE
 sklearn.tree._tree.TREE_UNDEFINED
-sklearn.tree._tree.Tree.children_left
-sklearn.tree._tree.Tree.children_right
 sklearn.tree._tree.Tree.compute_partial_dependence
-sklearn.tree._tree.Tree.feature
-sklearn.tree._tree.Tree.impurity
-sklearn.tree._tree.Tree.max_depth
-sklearn.tree._tree.Tree.missing_go_to_left
-sklearn.tree._tree.Tree.n_leaves
-sklearn.tree._tree.Tree.n_node_samples
-sklearn.tree._tree.Tree.threshold
-sklearn.tree._tree.Tree.weighted_n_node_samples
 sklearn.tree._tree.TreeBuilder.__reduce_cython__
 sklearn.tree._tree.TreeBuilder.__setstate_cython__
 sklearn.tree._tree.__pyx_capi__
@@ -1981,7 +1972,6 @@ sklearn.metrics._plot.precision_recall_curve.PrecisionRecallDisplay.plot
 sklearn.metrics._plot.roc_curve.RocCurveDisplay.from_estimator
 sklearn.metrics._plot.roc_curve.RocCurveDisplay.from_predictions
 sklearn.metrics._plot.roc_curve.RocCurveDisplay.plot
-sklearn.metrics._ranking.precision_recall_curve
 sklearn.metrics._regression.mean_squared_error
 sklearn.metrics._regression.mean_squared_log_error
 sklearn.metrics._scorer.SCORERS
@@ -2003,7 +1993,6 @@ sklearn.metrics.pairwise.check_pairwise_arrays
 sklearn.metrics.pairwise.manhattan_distances
 sklearn.metrics.pairwise.pairwise_distances
 sklearn.metrics.pairwise_distances
-sklearn.metrics.precision_recall_curve
 sklearn.model_selection.GridSearchCV.classes_
 sklearn.model_selection.GridSearchCV.n_features_in_
 sklearn.model_selection.GroupKFold.__init__

stubs/sklearn/tree/_tree.pyi

@@ -63,17 +63,36 @@ class BestFirstTreeBuilder(TreeBuilder):
 
 class Tree:
     n_features: int
-    n_classes: int
     n_outputs: int
     max_n_classes: int
 
-    max_dept: int
+    max_depth: int
     node_count: int
     capacity: int
     nodes: np.ndarray
     value: np.ndarray
     value_stride: int
 
+    @property
+    def n_classes(self) -> np.ndarray: ...
+    @property
+    def children_left(self) -> np.ndarray: ...
+    @property
+    def children_right(self) -> np.ndarray: ...
+    @property
+    def n_leaves(self) -> int: ...
+    @property
+    def feature(self) -> np.ndarray: ...
+    @property
+    def threshold(self) -> np.ndarray: ...
+    @property
+    def impurity(self) -> np.ndarray: ...
+    @property
+    def n_node_samples(self) -> np.ndarray: ...
+    @property
+    def weighted_n_node_samples(self) -> np.ndarray: ...
+    @property
+    def missing_go_to_left(self) -> np.ndarray: ...
     def predict(self, X) -> np.ndarray: ...
     def apply(self, X) -> np.ndarray: ...
     def decision_path(self, X) -> Any: ...