get_feature_names

pybear.base.get_feature_names(X)

Get feature names from X.

X must have a ‘columns’ attribute or a ‘__dataframe__’ dunder, i.e., follows the dataframe interchange protocol. Otherwise, feature names are not retrieved and None is returned. If the dataframe does not have a header comprised of strings (the dataframe was constructed without passing a header and a default non-string header is used), a warning is raised and None is returned.

Parameters:

Xarray_like of shape (n_samples, n_features) or (n_samples, )

Array container from which to extract feature names.

Objects that have known compatibility with this module: pandas dataframe and, polars dataframe. The columns will be considered to be feature names. If the dataframe contains non-string feature names, None is returned.

All other array containers will return ‘None’. Examples of containers known to not yield feature names: numpy arrays and scipy sparse matrices / arrays.

Returns:

feature_namesnumpy.ndarray[object] | None

The feature names of X. Unrecognized array containers return None.

Examples

>>> from pybear.base import get_feature_names
>>> import numpy as np
>>> import pandas as pd
>>> data = np.random.uniform(0, 1, (5, 3))
>>> X = pd.DataFrame(data=data, columns=['A', 'B', 'C'])
>>> out = get_feature_names(X)
>>> print(out)
['A' 'B' 'C']