sorbetto.flavor.correlation_flavor module

class sorbetto.flavor.correlation_flavor.CorrelationFlavor(performances: FiniteSetOfTwoClassClassificationPerformances, score: Callable[[TwoClassClassificationPerformance | FiniteSetOfTwoClassClassificationPerformances], ndarray], correlation_coefficient: Literal['pearson_r', 'spearman_rho', 'kendall_tau'] = 'pearson_r', name: str = 'Correlation Flavor', colormap: Any = None)[source]

Bases: AbstractNumericFlavor

For a given performance, the Correlation Flavor is the mathematical function that gives, to any importance \(I\) (that is, some application-specific preferences), the correlation, using a defined correlation coefficient (e.g., Pearson’s r), between a score \(X\) and the Ranking Score \(R_I\) corresponding to this importance.

property correlation_coefficient: str

getDefaultColormap()[source]

getLowerBound()[source]

getUpperBound()[source]

property performances: FiniteSetOfTwoClassClassificationPerformances

property score: Callable[[TwoClassClassificationPerformance | FiniteSetOfTwoClassClassificationPerformances], ndarray]