ModelProxy Class

Note

This is an experimental class, and may change at any time. Please see https://aka.ms/azuremlexperimental for more information.

Proxy object for AutoML models that enables inference on remote compute.

Create an AutoML ModelProxy object to submit inference to the training environment.

Inheritance

builtins.object

ModelProxy

Constructor

ModelProxy(child_run, compute_target=None)

Parameters

Name	Description
child_run Required	The child run from which the model will be downloaded.
compute_target Required	Overwrite for the target compute to inference on.

Methods

forecast	Submit a job to run forecast on the model for the given values.
forecast_quantiles	Submit a job to run forecast_quantiles on the model for the given values.
predict	Submit a job to run predict on the model for the given values.
predict_proba	Submit a job to run predict_proba on the model for the given values.
test	Retrieve predictions from the test_data and compute relevant metrics.

forecast

Submit a job to run forecast on the model for the given values.

forecast(X_values: Any, y_values: Any | None = None) -> Tuple[AbstractDataset, AbstractDataset]

Parameters

Name	Description
X_values Required	AbstractDataset or DataFrame or ndarray Input test data to run forecast on.
y_values	AbstractDataset or DataFrame or ndarray Input y values to run the forecast on. default value: None

Returns

Type	Description
	The forecast values.

forecast_quantiles

Submit a job to run forecast_quantiles on the model for the given values.

forecast_quantiles(X_values: Any, y_values: Any | None = None, forecast_destination: Any | None = None, ignore_data_errors: bool = False) -> AbstractDataset

Parameters

Name	Description
X_values Required	AbstractDataset Input test data to run forecast on.
y_values	Input y values to run the forecast on. default value: None
forecast_destination	<xref:pandas.Timestamp> Forecast_destination: a time-stamp value. Forecasts will be made all the way to the forecast_destination time, for all grains. Dictionary input { grain -> timestamp } will not be accepted. If forecast_destination is not given, it will be imputed as the last time occurring in X_pred for every grain. default value: None
ignore_data_errors	bool Ignore errors in user data. default value: False

predict

Submit a job to run predict on the model for the given values.

predict(values: Any) -> AbstractDataset

Parameters

Name	Description
values Required	AbstractDataset or DataFrame or ndarray Input test data to run predict on.

Returns

Type	Description
	The predicted values.

predict_proba

Submit a job to run predict_proba on the model for the given values.

predict_proba(values: Any) -> AbstractDataset

Parameters

Name	Description
values Required	AbstractDataset or DataFrame or ndarray Input test data to run predict on.

Returns

Type	Description
	The predicted values.

test

Retrieve predictions from the test_data and compute relevant metrics.

test(test_data: AbstractDataset, include_predictions_only: bool = False) -> Tuple[AbstractDataset, Dict[str, Any]]

Parameters

Name	Description
test_data Required	The test dataset.
include_predictions_only	Whether or not to only include the predictions as part of the predictions.csv output. If this parameter is True then the output CSV columns look like (forecasting is the same as regression): Classification => [predicted values], [probabilities] Regression => [predicted values] else (default): Classification => [original test data labels], [predicted values], [probabilities], [features] Regression => [original test data labels], [predicted values], [features] The [original test data labels] column name = [label column name] + "_orig". The [predicted values] column name = [label column name] + "_predicted". The [probabilities] column names = [class name] + "_predicted_proba". The [features] column names = [feature column name] + "_orig". If the test_data does not include a target column then [original test data labels] will not be in the output dataframe. default value: False

Returns

Type	Description
	A tuple containing the predicted values and the metrics.