nlptest.augmentation.AugmentRobustness#

class AugmentRobustness(task, h_report, config, max_prop=0.5)#

Bases: BaseAugmentaion

A class for performing a specified task with historical results.

task#

A string indicating the task being performed.

Type:

str

config#

A dictionary containing configuration parameters for the task.

Type:

dict

h_report#

A DataFrame containing a report of historical results for the task.

Type:

pandas.DataFrame

max_prop#

The maximum proportion of improvement that can be suggested by the class methods. Defaults to 0.5.

Type:

float

__init__(self, task, h_report, config, max_prop=0.5) → None#

Initializes an instance of MyClass with the specified parameters.

fix(self) → List[Sample]#

.

suggestions(self, prop) → pandas.DataFrame#

Calculates suggestions for improving test performance based on a given report.

__init__(task, h_report, config, max_prop=0.5) → None#

Initializes an instance of MyClass with the specified parameters.

Parameters:

  • task (str) – A string indicating the task being performed.

  • h_report (pandas.DataFrame) – A DataFrame containing a report of historical results for the task.

  • config (dict) – A dictionary containing configuration parameters for the task.

  • max_prop (float) – The maximum proportion of improvement that can be suggested by the class methods. Defaults to 0.5.

Returns:

None

Methods

__init__(task, h_report, config[, max_prop])

Initializes an instance of MyClass with the specified parameters.

fix(input_path, output_path[, inplace])

Applies perturbations to the input data based on the recommendations from harness reports.

suggestions(report)

Calculates suggestions for improving test performance based on a given report.
fix(input_path: str, output_path, inplace: bool = False)#

Applies perturbations to the input data based on the recommendations from harness reports.

Parameters:

  • input_path (str) – The path to the input data file.

  • output_path (str) – The path to save the augmented data file.

  • inplace (bool, optional) – If True, the list of samples is modified in place. Otherwise, a new samples are add to input data. Defaults to False.

Returns:

A list of augmented data samples.

Return type:

List[Dict[str, Any]]

suggestions(report)#

Calculates suggestions for improving test performance based on a given report.

Parameters:

report (pandas.DataFrame) – A DataFrame containing test results by category and test type, including pass rates and minimum pass rates.

Returns:

A DataFrame containing the following columns for each suggestion:

  • category: the test category

  • test_type: the type of test

  • ratio: the pass rate divided by the minimum pass rate for the test

  • proportion_increase: a proportion indicating how much the pass rate

    should increase to reach the minimum pass rate

Return type:

pandas.DataFrame