autogluon.cloud.TimeSeriesFoundationModel

class autogluon.cloud.TimeSeriesFoundationModel(model_id: str, **kwargs)[source]

Foundation model for time series forecasting (Chronos, etc.).

__init__(model_id: str, backend: Literal['sagemaker'] = 'sagemaker', cloud_output_path: str | None = None, hyperparameters: Dict[str, Any] | None = None, role: str | None = None)
Parameters:

  • model_id – ID of the foundation model from the model registry.

  • backend – Cloud backend to use.

  • cloud_output_path – S3 path to store intermediate artifacts.

  • hyperparameters – Default hyperparameters applied to inference and (when supported) training.

  • role – ARN of the SageMaker execution role used to run training and inference jobs. If None, falls back to role_arn in ~/.autogluon/cloud.yaml (set by autogluon.cloud.bootstrap() / autogluon.cloud.register()), and finally to sagemaker.get_execution_role().

deploy(instance_type: str | None = None, endpoint_name: str | None = None, hyperparameters: Dict[str, Any] | None = None, framework_version: str = 'latest', custom_image_uri: str | None = None, wait: bool = True, **backend_kwargs) → TimeSeriesEndpoint[source]

Deploy model to a real-time endpoint.

Parameters:

  • instance_type – Instance type for the endpoint. If None, will use the default from the model registry.

  • endpoint_name – Custom endpoint name. If None, will auto-generate a unique name.

  • hyperparameters – Model hyperparameters for inference. Overrides values passed to the constructor.

  • framework_version – Container framework version. If ‘latest’, uses the most recent available.

  • custom_image_uri – Custom Docker image URI for the inference container.

  • wait – Whether to block until the endpoint is ready.

  • **backend_kwargs – Backend-specific arguments (e.g., initial_instance_count, volume_size, model_kwargs, deploy_kwargs).

Return type:

TimeSeriesEndpoint

predict(data: str | DataFrame, target: str = 'target', id_column: str = 'item_id', timestamp_column: str = 'timestamp', known_covariates: str | DataFrame | None = None, static_features: str | DataFrame | None = None, prediction_length: int = 1, quantile_levels: List[float] | None = None, hyperparameters: Dict[str, Any] | None = None, instance_type: str | None = None, framework_version: str = 'latest', custom_image_uri: str | None = None, wait: bool = True, **backend_kwargs) → DataFrame | None[source]

Run batch prediction for time series.

Parameters:

  • data – Historical time series in long format (DataFrame or S3 path).

  • target – Name of the target column to forecast.

  • id_column – Name of the item ID column.

  • timestamp_column – Name of the timestamp column.

  • known_covariates – Future values of known covariates (DataFrame or S3 path). Covariate column names are inferred from the DataFrame columns (excluding id_column and timestamp_column).

  • static_features – Metadata attributes of individual items (DataFrame or S3 path).

  • prediction_length – Number of time steps to forecast.

  • quantile_levels – Quantiles to predict.

  • hyperparameters – Model hyperparameters for inference. Overrides values passed to the constructor.

  • instance_type – Instance type for the prediction job. If None, uses registry default.

  • framework_version – Container framework version.

  • custom_image_uri – Custom Docker image URI for the container.

  • wait – If True, block and return DataFrame. If False, return the job handle.

  • **backend_kwargs – Additional backend-specific arguments (e.g., job_name, volume_size, autogluon_sagemaker_estimator_kwargs).

Return type:

Optional[pd.DataFrame]