feature_name_mapper

pybear.utilities.feature_name_mapper(feature_names, feature_names_in, positive=True)

Map a vector of feature names feature_names against the full set of feature names provided by feature_names_in.

Return the index positions of the given feature names with respect to their position in feature_names_in. Can be returned as positive indices, e.g. [0,3,4], or negative indices, e.g. [-4,-2,-1], or as given.

Parameters:

feature_namesSequence[str] | Sequence[int] | None

The feature names to be mapped to index positions. If None, returns None. If an empty 1D sequence, returns the same. If passed as integers without a feature_names_in reference, returns the original. if passed as integers with a feature_names_in reference, the index values are validated against the dimensions of the feature_names_in vector and mapped to all positive or all negative values based on positive. If passed as strings without a feature_names_in reference, raises exception. If passed as strings with a feature_names_in reference, the string values are mapped to the index positions in the feature_names_in vector and mapped to all positive or all negative values based on positive. If passed as string values and a value is not in feature_names_in, raises exception.

feature_names_inSequence[str] | None

If not None, a 1D list-like containing strings that are the feature names of a data-bearing container.

positivebool | None

Whether to return the mapped indices as all positive or all negative integers. if None, leave the indices as is.

Returns:

indicesnumpy.ndarray[np.int32]

The given feature names mapped to index positions.

Examples

>>> from pybear.utilities import feature_name_mapper
>>> import numpy as np
>>> data = np.random.randint(0, 10, (5, 3))
>>> columns = np.array(['A', 'B', 'C'])
>>> feature_names = np.array(['A', 'C'])
>>> out = feature_name_mapper(
...     feature_names, columns, positive=False
... )
>>> print(out)
[-3 -1]