Relabel#

class sdm.prior.tabular.transform.Relabel(random_function: RandomFunction, *, exact_probability: float = 0.5)#

Bases: PriorTransform

Recompute the target from selected feature columns.

Equal inputs receive equal scores. Classification retains the input target when the scores cannot separate the declared classes.

Parameters:
  • random_function (RandomFunction) – Function used to score the selected features.

  • exact_probability (float) – Probability that classification uses exact score cuts rather than a noisy score rank.