citrine.informatics.predictor_evaluation_metrics module

class citrine.informatics.predictor_evaluation_metrics.AreaUnderROC

Bases: Serializable[AreaUnderROC], PredictorEvaluationMetric

Area under the receiver operating characteristic (ROC) curve.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'AreaUnderROC'

class citrine.informatics.predictor_evaluation_metrics.CoverageProbability(*, coverage_level: str | float = '0.683')

Bases: Serializable[CoverageProbability], PredictorEvaluationMetric

Percentage of observations that fall within a given confidence interval.

The coverage level can be specified to 3 digits, e.g., 0.123, but not 0.1234.

Parameters:

coverage_level (Union[str, float]) – Confidence-interval coverage level. The coverage level must be between 0 and 1.0 (non-inclusive) and will be rounded to 3 significant figures. Default: 0.683 corresponds to one std. deviation

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'CoverageProbability'

class citrine.informatics.predictor_evaluation_metrics.F1

Bases: Serializable[F1], PredictorEvaluationMetric

Support-weighted F1 score.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'F1'

class citrine.informatics.predictor_evaluation_metrics.NDME

Bases: Serializable[NDME], PredictorEvaluationMetric

Non-dimensional model error.

The non-dimensional model error is the RMSE divided by the standard deviation of the labels in the training data (including all folds, not just the training folds).

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'NDME'

class citrine.informatics.predictor_evaluation_metrics.PVA

Bases: Serializable[PVA], PredictorEvaluationMetric

Predicted vs. actual data.

Results are returned as a flattened list, where each item represents predicted vs. actual data for a single point.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'PVA'

class citrine.informatics.predictor_evaluation_metrics.PredictorEvaluationMetric

Bases: PolymorphicSerializable[PredictorEvaluationMetric]

A metric computed during a Predictor Evaluation Workflow.

Abstract type that returns the proper type given a serialized dict.

classmethod build(data: dict) → SelfType

Build the underlying type.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

class citrine.informatics.predictor_evaluation_metrics.RMSE

Bases: Serializable[RMSE], PredictorEvaluationMetric

Root-mean-square error.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'RMSE'

class citrine.informatics.predictor_evaluation_metrics.RSquared

Bases: Serializable[RSquared], PredictorEvaluationMetric

Fraction of variance explained, commonly known as R^2.

This dimensionless metric is equal to 1 - (mean squared error / variance of data). It is equal to the coefficient of determination calculated with respect to the line predicted = actual, hence it is commonly referred to as R^2. But unlike R^2 from ordinary linear regression, this metric can be negative.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'RSquared'

class citrine.informatics.predictor_evaluation_metrics.StandardRMSE

Bases: Serializable[StandardRMSE], PredictorEvaluationMetric

Standardized root-mean-square error.

classmethod build(data: dict) → Self

Build an instance of this object from given data.

dump() → dict

Dump this instance.

classmethod get_type(data) → Type[Serializable]

Return the subtype.

typ = 'StandardRMSE'