climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core package

Submodules

climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.confidence module

class climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.confidence.ConfidencePredictor(calibrator=None)

Bases: PerfPredictor

fit(x_test_unprocessed, x_test, y_test)

get_conf_dict(top_confidences, y_test)

load(input_location)

classmethod name()

Name of this subtype, used for lookup purposes. To be implemented by subclasses. This method can either return a single name, or a list/tuple of names.

predict(X_unprocessed, X)

save(output_location)

climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.passthrough module

class climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.passthrough.PassthroughPredictor(calibrator=None)

Bases: PerfPredictor

fit(x_test_unprocessed, x_test, y_test)

load(input_location)

classmethod name()

Name of this subtype, used for lookup purposes. To be implemented by subclasses. This method can either return a single name, or a list/tuple of names.

predict(X_unprocessed, X)

save(output_location)

climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.short_text module

class climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.short_text.TextEnsemblePredictor(calibrator='shift')

Bases: PerfPredictor

fit(x_test_unprocessed, x_test, y_test)

load(input_location)

classmethod name()

Name of this subtype, used for lookup purposes. To be implemented by subclasses. This method can either return a single name, or a list/tuple of names.

predict(X_unprocessed, X)

save(output_location)

climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.short_text.logger = <Logger climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.short_text (INFO)>

Performance predictor for short text data. It is based on an ensemble of meta-models: one mlp metamodel, one GBM metamodel, and one SVM metamodel. This performance predictor does not have a method to quantify its own uncertainty, so the uncertainty values are zero.

climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.structured_data module

class climb.tool.impl.data_suite.third_party.uq360.algorithms.blackbox_metamodel.predictors.core.structured_data.StructuredDataPredictor(calibrator=None)

Bases: PerfPredictor

fit(x_test_unprocessed, x_test, y_test)

init_all_whitebox_features()

load(input_location)

classmethod name()

Name of this subtype, used for lookup purposes. To be implemented by subclasses. This method can either return a single name, or a list/tuple of names.

predict(X_unprocessed, X)

save(output_location)

Module contents