agreement_numerical()

audpsychometric.agreement_numerical(ratings, minimum, maximum, *, axis=1)

Confidence score for numerical ratings.

\[\text{agreement}(\text{ratings}) = \max( 0, 1 - \frac{\text{std}(\text{ratings})} {\text{maximum} - \frac{1}{2} (\text{minimum} + \text{maximum})} )\]

with \(\text{std}\) the population standard deviation of the ratings. nan values are ignored per item.

Parameters

• ratings (Sequence) – ratings. When given as a 1-dimensional array, it is treated as a row vector

• minimum (float) – lower limit of possible rating value

• maximum (float) – upper limit of possible rating value

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

Return type

Union[float, ndarray]

Returns

numerical agreement score(s)

Examples

>>> agreement_numerical([0, 1], 0, 1)
0.0
>>> agreement_numerical([0, 1], 0, 2)
0.5
>>> agreement_numerical([0, 0], 0, 1)
1.0
>>> agreement_numerical([0, np.nan], 0, 1)
1.0