-
Notifications
You must be signed in to change notification settings - Fork 102
Open
Description
In sklearn 1.8, the "enforce" arguments of check_X_y and check_array have been changed to "ensure" in order to reach a more coherent notation
"enforce" was deprecated as of 1.7, and has been dropped in 1.8.
As of now, the stub is :
def check_X_y(
X: MatrixLike | ArrayLike,
y: MatrixLike | ArrayLike,
accept_sparse: Sequence[str] | tuple[str, str] | list[str] | str | bool = False,
*,
accept_large_sparse: bool = True,
dtype: None | Sequence[type] | Literal["numeric"] | type = "numeric",
order: Literal["F", "C"] | None = None,
copy: bool = False,
force_all_finite: str | bool = True, ### This is the offending line
ensure_2d: bool = True,
allow_nd: bool = False,
multi_output: bool = False,
ensure_min_samples: Int = 1,
ensure_min_features: Int = 1,
y_numeric: bool = False,
estimator: None | str | BaseEstimator = None,
) -> tuple[Any, Any]: ...When it should really be (for sklearn >=1.8) :
def check_X_y(
X: MatrixLike | ArrayLike,
y: MatrixLike | ArrayLike,
accept_sparse: Sequence[str] | tuple[str, str] | list[str] | str | bool = False,
*,
accept_large_sparse: bool = True,
dtype: None | Sequence[type] | Literal["numeric"] | type = "numeric",
order: Literal["F", "C"] | None = None,
copy: bool = False,
ensure_all_finite: str | bool = True, ### This is the corrected line
ensure_2d: bool = True,
allow_nd: bool = False,
multi_output: bool = False,
ensure_min_samples: Int = 1,
ensure_min_features: Int = 1,
y_numeric: bool = False,
estimator: None | str | BaseEstimator = None,
) -> tuple[Any, Any]: ...Note : Same applies for check_array
This only applies from 1.8 and onwards
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels