kumoai.encoder.Index#

class kumoai.encoder.Index[source]#

Bases: Encoder

An Index encoder encodes its corresponding categorical column by assigning each unique value with frequency above min_occ to an embedding of size channels from the model plan. Values below this frequency are all collapsed to the same embedding.

min_occ: PositiveInt = 1#

The minimum frequency of distinct values.

na_strategy: Literal[zero, separate, most_frequent, raise] = 'separate'#

The specified null value imputation strategy.

__init__(min_occ=1, na_strategy=NAStrategy.SEPARATE, name='Index', _stats=<factory>, _target_='kumo.encoder.categorical.Index')#