from abc import ABC, abstractmethod
from collections.abc import Mapping, Sequence
from typing import Callable, ClassVar, Literal
from typing_extensions import Self

from numpy import ndarray
from numpy.random import RandomState

from ..._typing import ArrayLike, Float, Int, MatrixLike
from ...base import BaseEstimator, ClassifierMixin, RegressorMixin

class BaseHistGradientBoosting(BaseEstimator, ABC):
    _parameter_constraints: ClassVar[dict] = ...

    @abstractmethod
    def __init__(
        self,
        loss: str,
        *,
        learning_rate,
        max_iter,
        max_leaf_nodes,
        max_depth,
        min_samples_leaf,
        l2_regularization,
        max_features,
        max_bins,
        categorical_features,
        monotonic_cst,
        interaction_cst,
        warm_start,
        early_stopping,
        scoring,
        validation_fraction,
        n_iter_no_change,
        tol,
        verbose,
        random_state,
    ) -> None: ...
    def fit(
        self,
        X: MatrixLike,
        y: ArrayLike,
        sample_weight: None | ArrayLike = None,
    ) -> Self: ...
    def n_iter_(self): ...

class HistGradientBoostingRegressor(RegressorMixin, BaseHistGradientBoosting):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    is_categorical_: None | ndarray = ...
    validation_score_: ndarray = ...
    train_score_: ndarray = ...
    n_trees_per_iteration_: int = ...
    n_iter_: int = ...
    do_early_stopping_: bool = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(
        self,
        loss: Literal["squared_error", "absolute_error", "poisson", "quantile"] = "squared_error",
        *,
        quantile: None | Float = None,
        learning_rate: Float = 0.1,
        max_iter: Int = 100,
        max_leaf_nodes: None | int = 31,
        max_depth: None | int = None,
        min_samples_leaf: Int = 20,
        l2_regularization: Float = 0.0,
        max_features: Float = 1.0,
        max_bins: Int = 255,
        categorical_features: None | MatrixLike | ArrayLike = None,
        monotonic_cst: None | Mapping | ArrayLike = None,
        interaction_cst: None | Literal["pairwise", "no_interaction"] | Sequence[list[int] | tuple[int, ...] | set[int]] = None,
        warm_start: bool = False,
        early_stopping: Literal["auto"] | bool = "auto",
        scoring: None | str | Callable = "loss",
        validation_fraction: float | None = 0.1,
        n_iter_no_change: Int = 10,
        tol: Float = 1e-7,
        verbose: Int = 0,
        random_state: RandomState | None | Int = None,
    ) -> None: ...
    def predict(self, X: MatrixLike) -> ndarray: ...
    def staged_predict(self, X: MatrixLike): ...

class HistGradientBoostingClassifier(ClassifierMixin, BaseHistGradientBoosting):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    is_categorical_: None | ndarray = ...
    validation_score_: ndarray = ...
    train_score_: ndarray = ...
    n_trees_per_iteration_: int = ...
    n_iter_: int = ...
    do_early_stopping_: bool = ...
    classes_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(
        self,
        loss: Literal["log_loss", "auto", "binary_crossentropy", "categorical_crossentropy"] = "log_loss",
        *,
        learning_rate: Float = 0.1,
        max_iter: Int = 100,
        max_leaf_nodes: None | int = 31,
        max_depth: None | int = None,
        min_samples_leaf: Int = 20,
        l2_regularization: Float = 0.0,
        max_features: Float = 1.0,
        max_bins: Int = 255,
        categorical_features: None | MatrixLike | ArrayLike = None,
        monotonic_cst: None | Mapping | ArrayLike = None,
        interaction_cst: None | Literal["pairwise", "no_interaction"] | Sequence[list[int] | tuple[int, ...] | set[int]] = None,
        warm_start: bool = False,
        early_stopping: Literal["auto"] | bool = "auto",
        scoring: None | str | Callable = "loss",
        validation_fraction: float | None = 0.1,
        n_iter_no_change: Int = 10,
        tol: Float = 1e-7,
        verbose: Int = 0,
        random_state: RandomState | None | Int = None,
        class_weight: None | Mapping | str = None,
    ) -> None: ...
    def predict(self, X: MatrixLike) -> ndarray: ...
    def staged_predict(self, X: MatrixLike): ...
    def predict_proba(self, X: MatrixLike) -> ndarray: ...
    def staged_predict_proba(self, X: MatrixLike): ...
    def decision_function(self, X: MatrixLike) -> ndarray: ...
    def staged_decision_function(self, X: MatrixLike): ...