skopt.space.space.check_dimension

skopt.space.space.check_dimension(dimension, transform=None)

Turn a provided dimension description into a dimension object.

Checks that the provided dimension falls into one of the supported types. For a list of supported types, look at the documentation of dimension below.

If dimension is already a Dimension instance, return it.

Parameters:

dimensionDimension

Search space Dimension. Each search dimension can be defined either as:

• an instance of a Dimension object (Real, Integer or Categorical).

• a 2-, 3- or 4-tuple, for Real and Integer dimensions, of the form (low, high [, prior [, base]]) (values in square brackets are optional). If both low and high are integral numbers (as per the number.Integral), a Integer dimension is returned, else a Real dimension is returned.

• any iterable for Categorical dimension

  Note

  For a transitionary period, the old behavior is retained. This means tuple, list and array currently all undergo dimension inference as describe in the tuple entry above. If no Integer or Real dimension can be inferred, a Categorical is returned. This behavior will be tightened to the above description in an upcoming version, and a warning is raised if the upcoming inference would differ from the current behavior.

transform“identity”, “normalize”, “string”, “label”, “onehot” optional

• For Categorical dimensions, the following transformations are supported.

  • “onehot” (default) one-hot transformation of the original space.

  • “label” integer transformation of the original space

  • “string” string transformation of the original space.

  • “identity” same as the original space.

• For Real and Integer dimensions, the following transformations are supported.

  • “identity”, (default) the transformed space is the same as the original space.

  • “normalize”, the transformed space is scaled to be between 0 and 1.

Returns:

dimensionDimension

Dimension instance.