verticapy.mlops.model_versioning.RegisteredModel.predict_proba¶
- RegisteredModel.predict_proba(vdf: Annotated[str | vDataFrame, ''], X: Annotated[str | list[str], 'STRING representing one column or a list of columns'] | None = None, name: str | None = None, pos_label: Annotated[bool | float | str | timedelta | datetime, 'Python Scalar'] | None = None, inplace: bool = True, version: int | None = None) vDataFrame¶
Returns the model’s probabilities using the input relation.
Parameters¶
- vdf: SQLRelation
Object used to run the prediction. You can also specify a customized relation, but you must enclose it with an alias. For example,
(SELECT 1) xis valid, whereas(SELECT 1)and “SELECT 1” are invalid.- X: SQLColumns, optional
List of the columns used to deploy the models. If empty, the model predictors are used.
- name: str, optional
Name of the added vDataColumn. If empty, a name is generated.
- pos_label: PythonScalar, optional
Class label. For binary classification, this can be either 1 or 0.
- inplace: bool, optional
If set to True, the prediction is added to the vDataFrame.
- version: int, optional
When the version is None, the registered model with “production” status will be used for prediction. When the version is specified, the registered model that version will be used. It will throw an error if it doesn’t find such a model.
Returns¶
- vDataFrame
the input object.