RecodeHighCardinality#

class sdm.prior.tabular.transform.RecodeHighCardinality(*, min_levels: int = 201, num_columns: tuple[int, int] = (1, 3), categorical: bool = False, standardize: bool = False)#

Bases: PriorTransform

Recode selected finite feature columns as high-cardinality categories.

Parameters:
  • min_levels (int) – Minimum requested levels and required number of rows.

  • num_columns (tuple[int, int]) – Inclusive range for the number of columns to recode.

  • categorical (bool) – Whether existing categorical columns are eligible.

  • standardize (bool) – Whether category values use standardized numeric levels.