from abc import ABCMeta, abstractmethod
from collections.abc import Iterator
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
from ._base import BaseEnsemble
from ._gb_losses import LossFunction

class VerboseReporter:
    def __init__(self, verbose: Int) -> None: ...
    def init(self, est: BaseEstimator, begin_at_stage: Int = 0): ...
    def update(self, j: Int, est: BaseEstimator): ...

class BaseGradientBoosting(BaseEnsemble, metaclass=ABCMeta):
    _parameter_constraints: ClassVar[dict] = ...

    @abstractmethod
    def __init__(
        self,
        *,
        loss,
        learning_rate,
        n_estimators,
        criterion,
        min_samples_split,
        min_samples_leaf,
        min_weight_fraction_leaf,
        max_depth,
        min_impurity_decrease,
        init,
        subsample,
        max_features,
        ccp_alpha,
        random_state,
        alpha: float = 0.9,
        verbose: int = 0,
        max_leaf_nodes=None,
        warm_start: bool = False,
        validation_fraction: float = 0.1,
        n_iter_no_change=None,
        tol: float = 1e-4,
    ) -> None: ...
    def fit(
        self,
        X: MatrixLike | ArrayLike,
        y: ArrayLike,
        sample_weight: None | ArrayLike = None,
        monitor: None | Callable = None,
    ) -> Self: ...
    @property
    def feature_importances_(self) -> ndarray: ...
    def apply(self, X: MatrixLike | ArrayLike) -> ndarray: ...

class GradientBoostingClassifier(ClassifierMixin, BaseGradientBoosting):
    max_features_: int = ...
    n_classes_: int = ...
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    classes_: ndarray = ...
    estimators_: ndarray = ...
    init_: BaseEstimator = ...
    loss_: LossFunction = ...
    train_score_: ndarray = ...
    oob_improvement_: ndarray = ...
    feature_importances_: ndarray = ...
    n_estimators_: int = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(
        self,
        *,
        loss: Literal["log_loss", "deviance", "exponential"] = "log_loss",
        learning_rate: Float = 0.1,
        n_estimators: Int = 100,
        subsample: Float = 1.0,
        criterion: Literal["friedman_mse", "squared_error"] = "friedman_mse",
        min_samples_split: float = 2,
        min_samples_leaf: float = 1,
        min_weight_fraction_leaf: Float = 0.0,
        max_depth: None | int = 3,
        min_impurity_decrease: Float = 0.0,
        init: None | str | BaseEstimator = None,
        random_state: RandomState | None | Int = None,
        max_features: float | None | Literal["auto", "sqrt", "log2"] = None,
        verbose: Int = 0,
        max_leaf_nodes: None | Int = None,
        warm_start: bool = False,
        validation_fraction: Float = 0.1,
        n_iter_no_change: None | Int = None,
        tol: Float = 1e-4,
        ccp_alpha: float = 0.0,
    ) -> None: ...
    def decision_function(self, X: MatrixLike | ArrayLike) -> ndarray: ...
    def staged_decision_function(self, X: MatrixLike | ArrayLike): ...
    def predict(self, X: MatrixLike | ArrayLike) -> ndarray: ...
    def staged_predict(self, X: MatrixLike | ArrayLike): ...
    def predict_proba(self, X: MatrixLike | ArrayLike) -> ndarray: ...
    def predict_log_proba(self, X: MatrixLike | ArrayLike) -> ndarray: ...
    def staged_predict_proba(self, X: MatrixLike | ArrayLike) -> Iterator[ndarray]: ...

class GradientBoostingRegressor(RegressorMixin, BaseGradientBoosting):
    max_features_: int = ...
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    n_estimators_: int = ...
    estimators_: ndarray = ...
    init_: BaseEstimator = ...
    loss_: LossFunction = ...
    train_score_: ndarray = ...
    oob_improvement_: ndarray = ...
    feature_importances_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(
        self,
        *,
        loss: Literal["squared_error", "absolute_error", "huber", "quantile"] = "squared_error",
        learning_rate: Float = 0.1,
        n_estimators: Int = 100,
        subsample: Float = 1.0,
        criterion: Literal["friedman_mse", "squared_error"] = "friedman_mse",
        min_samples_split: float = 2,
        min_samples_leaf: float = 1,
        min_weight_fraction_leaf: Float = 0.0,
        max_depth: None | int = 3,
        min_impurity_decrease: Float = 0.0,
        init: None | str | BaseEstimator = None,
        random_state: RandomState | None | Int = None,
        max_features: float | None | Literal["auto", "sqrt", "log2"] = None,
        alpha: Float = 0.9,
        verbose: Int = 0,
        max_leaf_nodes: None | Int = None,
        warm_start: bool = False,
        validation_fraction: Float = 0.1,
        n_iter_no_change: None | Int = None,
        tol: Float = 1e-4,
        ccp_alpha: float = 0.0,
    ) -> None: ...
    def predict(self, X: MatrixLike | ArrayLike) -> ndarray: ...
    def staged_predict(self, X: MatrixLike | ArrayLike) -> Iterator[ndarray]: ...
    def apply(self, X: MatrixLike | ArrayLike) -> ndarray: ...