utilities Module

Utility methods for validation and conversion.

Classes

suppress_stdout_stderr

A context manager for doing a "deep suppression" of stdout and stderr. Will suppress all print, even if the print originates in a compiled C/Fortran sub-function. Will not suppress raised exceptions, since exceptions are printed to stderr just before a script exits, and after the context manager has exited. Create the context manager.

Functions

convert_dict_values_to_str

Convert a dictionary's values so that every value is a string.

convert_dict_values_to_str(input_dict: Dict[Any, Any]) -> Dict[str, str]

Parameters

Name	Description
input_dict Required	the dictionary that should be converted

Returns

Type	Description
	a dictionary with all values converted to strings

get_default_metric_with_objective

Get the dictionary of metric -> objective for the given task.

get_default_metric_with_objective(task)

Parameters

Name	Description
task Required	string "classification" or "regression"

Returns

Type	Description
	dictionary of metric -> objective

get_error_code

Build the error code from an exception.

get_error_code(exception: BaseException, as_hierarchy: bool = False) -> str

Parameters

Name	Description
exception Required	The exception that fails the run.
as_hierarchy	If the complete error hierarchy should be returned Default value: False

Returns

Type	Description
	Return the str containing error_code. If as_hierarchy is True, the hierarchy returned is joined by a '.'

get_min_points

Return the minimum number of data points needed for training.

get_min_points(window_size: int, lags: List[int], max_horizon: int, cv: int | None, n_step: int | None = None) -> int

Parameters

Name	Description
window_size Required	the rolling window size.
lags Required	The lag size.
max_horizon Required	the desired length of forecasting.
cv Required	the number of cross validations.
n_step	Number of periods between the origin_time of one CV fold and the next fold. For example, if n_step = 3 for daily data, the origin time for each fold will be three days apart. Default value: None

Returns

Type	Description
	the minimum number of data points.

get_primary_metrics

Get the primary metrics supported for a given task as a list.

get_primary_metrics(task: str) -> List[str]

Parameters

Name	Description
task Required	str Task type supported by AutoML, as defined in azureml.automl.core.shared.constants.Tasks

Returns

Type	Description
	A list of the primary metrics supported for the task.

get_value_float

Convert string value to float. :param floatstring: The input value to be converted. :type floatstring: str :return: The converted value. :rtype: float

get_value_float(floatstring: str) -> float | str | None

Parameters

Name	Description
floatstring Required

get_value_from_dict

Get the value of a configuration item that has a list of names.

get_value_from_dict(dictionary: Dict[str, Any], names: List[str], default_value: Any) -> Any

Parameters

Name	Description
dictionary Required	dict Dictionary of settings with key value pair to look the data for.
names Required	list[str] The list of names for the item looking foi.
default_value Required	Default value to return if no matching key found

Returns

Type	Description
	Returns the first value from the list of names.

get_value_int

Convert string value to int.

get_value_int(intstring: str) -> int | str | None

Parameters

Name	Description
intstring Required	str The input value to be converted.

Returns

Type	Description
int	The converted value.

interpret_exception

Translate an exception to an AzureMLException.

If the exception is already one of the known types (e.g. ServiceException, AzureMLException), return the exception as-is.

Dev note: If we see adding more exceptions, or new interpretations for remote vs. local runs, consider converting this functionality into its own class

interpret_exception(exception: BaseException, is_aml_compute: bool = True, **kwargs: Any) -> AzureMLException | ServiceException

Parameters

Name	Description
exception Required	The exception object that needs to be interpreted
is_aml_compute	If the context is an execution service managed run on an AML Compute (e.g. OSErrors, networking errors may need to be interpreted differently based on the run type) Default value: True
kwargs Required	Any run-time properties that the ErrorDefinition expects (such as reference_code)

Returns

Type	Description
	exception interpreted as an AzureMLException with error code

is_known_date_time_format

Check if a given string matches the known date time regular expressions.

is_known_date_time_format(datetime_str: str) -> bool

Parameters

Name	Description
datetime_str Required	Input string to check if it's a date or not

Returns

Type	Description
	Whether the given string is in a known date time format or not

minimize_or_maximize

Select the objective given a metric.

Some metrics should be minimized and some should be maximized :param metric: the name of the metric to look up :param task: one of constants.Tasks. :return: returns one of constants.OptimizerObjectives.

minimize_or_maximize(metric, task=None)

Parameters

Name	Description
metric Required
task	Default value: None

subsampling_recommended

subsampling_recommended(num_samples)

Parameters

Name	Description
num_samples Required	int number of samples.

Returns

Type	Description
bool	True if subsampling is recommended, else False.

to_ordinal_string

Convert an integer to an ordinal string.

to_ordinal_string(integer: int) -> str

Parameters

Name	Description
integer Required