predict_proba

TabularEndpoint.predict_proba(data: str | Path | DataFrame, train_data: str | Path | DataFrame, label: str, *, include_predict: bool = True, **inference_kwargs: Any) → Tuple[Series, DataFrame | Series] | DataFrame | Series

Fit the foundation model and return class probabilities.

For regression, the probability result is identical to the prediction.

The serialized request includes both train_data and data and must not exceed SageMaker’s 6 MiB real-time invocation payload limit. Use autogluon.cloud.TabularFoundationModel.predict_proba() for larger inputs.