logml.metrics.registry.survival

Functions

calculate_concordance_index(y_true, y_score)

param y_true Survival target, np.array with structured dtype (censored, time)

Classes

CIndex(**kwargs)

c-index survival metric class.

logml.metrics.registry.survival.calculate_concordance_index(y_true: numpy.array, y_score: numpy.array, raise_error=False, **_kwargs)

Parameters

• y_true – Survival target, np.array with structured dtype (censored, time)

• y_score – Survival prediction (risk score).

Return type

float - ‘cindex’ from sksurv.metrics.concordance_index_censored

class logml.metrics.registry.survival.CIndex(**kwargs)

Bases: logml.metrics.base.BaseMetric

c-index survival metric class. Calculates alternative cindex metric as per Harrel.

LABEL: str = 'cindex'

TASK: logml.common.ModelingTask = 'survival'

get_value(y_true: Optional[numpy.ndarray] = None, y_pred: Optional[numpy.ndarray] = None, y_pred_proba: Optional[numpy.ndarray] = None, class_labels: Optional[list] = None, **kwargs)

Returns

float - ‘cindex’ from sksurv.metrics.concordance_index_censored