Categorical

CategoricalParameter (BaseParameter)

Categorical parameter that can take discrete values of any type.

Source code in parameterspace/parameters/categorical.py

class CategoricalParameter(BaseParameter):
    """Categorical parameter that can take discrete values of any type."""

    @store_init_arguments
    # pylint: disable-next=too-many-positional-arguments
    def __init__(
        self,
        name: str,
        values: Tuple[Any, ...],
        prior: Union[list, np.ndarray, None] = None,
        transformation: Optional[BaseTransformation] = None,
        inactive_numerical_value: Optional[float] = np.nan,
    ):
        """
        Initialize with options for categorical parameter.

        Args:
            name: Name of the parameter.
            values: Allowed values for this parameter.
            prior: Probabilities for each value (does not need to be normalized).
            transformation: A transformation that can translate the arbitrary type of
                the values to a numerical value. The only supported one right now
                is [parameterspace.transformations.categorical.Cat2Num][], which is
                also used as default.
            inactive_numerical_value: [description]
        """
        self.values = values
        transformation = Cat2Num(values) if transformation is None else transformation

        if prior is None:
            _prior = Categorical([1.0] * len(values))
        elif not isinstance(prior, Categorical):
            _prior = Categorical(prior)
        else:
            _prior = prior

        super().__init__(
            name,
            _prior,
            transformation,
            is_continuous=False,
            is_ordered=False,
            num_values=len(values),
            inactive_numerical_value=inactive_numerical_value,
        )

    def check_value(self, value):
        """Check if value is valid."""
        return value in self.values

__init__(self, name, values, prior=None, transformation=None, inactive_numerical_value=nan) special

Initialize with options for categorical parameter.

Parameters:

Name	Type	Description	Default
name	str	Name of the parameter.	required
values	Tuple[Any, ...]	Allowed values for this parameter.	required
prior	Union[list, numpy.ndarray]	Probabilities for each value (does not need to be normalized).	None
transformation	Optional[parameterspace.transformations.base.BaseTransformation]	A transformation that can translate the arbitrary type of the values to a numerical value. The only supported one right now is parameterspace.transformations.categorical.Cat2Num, which is also used as default.	None
inactive_numerical_value	Optional[float]	[description]	nan

Source code in parameterspace/parameters/categorical.py

@store_init_arguments
# pylint: disable-next=too-many-positional-arguments
def __init__(
    self,
    name: str,
    values: Tuple[Any, ...],
    prior: Union[list, np.ndarray, None] = None,
    transformation: Optional[BaseTransformation] = None,
    inactive_numerical_value: Optional[float] = np.nan,
):
    """
    Initialize with options for categorical parameter.

    Args:
        name: Name of the parameter.
        values: Allowed values for this parameter.
        prior: Probabilities for each value (does not need to be normalized).
        transformation: A transformation that can translate the arbitrary type of
            the values to a numerical value. The only supported one right now
            is [parameterspace.transformations.categorical.Cat2Num][], which is
            also used as default.
        inactive_numerical_value: [description]
    """
    self.values = values
    transformation = Cat2Num(values) if transformation is None else transformation

    if prior is None:
        _prior = Categorical([1.0] * len(values))
    elif not isinstance(prior, Categorical):
        _prior = Categorical(prior)
    else:
        _prior = prior

    super().__init__(
        name,
        _prior,
        transformation,
        is_continuous=False,
        is_ordered=False,
        num_values=len(values),
        inactive_numerical_value=inactive_numerical_value,
    )

check_value(self, value)

Check if value is valid.

Source code in parameterspace/parameters/categorical.py

def check_value(self, value):
    """Check if value is valid."""
    return value in self.values