Class CategoricalTransformation (1.36.4)

CategoricalTransformation(mapping=None, *, ignore_unknown_fields=False, **kwargs)

Training pipeline will perform following transformation functions.

  • The categorical string as is--no change to case, punctuation, spelling, tense, and so on.
  • Convert the category name to a dictionary lookup index and generate an embedding for each index.
  • Categories that appear less than 5 times in the training dataset are treated as the "unknown" category. The "unknown" category gets its own special lookup index and resulting embedding.

Methods

CategoricalTransformation

CategoricalTransformation(mapping=None, *, ignore_unknown_fields=False, **kwargs)

Training pipeline will perform following transformation functions.

  • The categorical string as is--no change to case, punctuation, spelling, tense, and so on.
  • Convert the category name to a dictionary lookup index and generate an embedding for each index.
  • Categories that appear less than 5 times in the training dataset are treated as the "unknown" category. The "unknown" category gets its own special lookup index and resulting embedding.

CategoricalTransformation

CategoricalTransformation(mapping=None, *, ignore_unknown_fields=False, **kwargs)

Training pipeline will perform following transformation functions.

  • The categorical string as is--no change to case, punctuation, spelling, tense, and so on.
  • Convert the category name to a dictionary lookup index and generate an embedding for each index.
  • Categories that appear less than 5 times in the training dataset are treated as the "unknown" category. The "unknown" category gets its own special lookup index and resulting embedding.