evaluator_weighted_estimator()

audpsychometric.evaluator_weighted_estimator(ratings, *, axis=1)[source]

Evaluator weighted estimator (EWE) of raters’ votes.

The EWE is described in [CHonigZ+16] as follows:

The EWE average of the individual ratings considers that each evaluator is subject to an individual amount of disturbance during the evaluation, by introducing evaluator-dependent weights that correspond to the correlation between the listener’s responses and the average ratings of all evaluators.

See also audformat#102 for implementation details.

Parameters

  • ratings (Sequence) – ratings. Has to contain more than one rater and more than one stimuli

  • axis (int) – axis along which the EWE is computed. A value of 1 assumes stimuli as rows and raters as columns

Return type

Union[float, ndarray]

Returns

EWE over raters

Examples

>>> evaluator_weighted_estimator([[1, 1, 0], [2, 2, 1]])
array([0.66666667, 1.66666667])