subrela.records.from_arrays function

subrela.records.from_arrays(flags, scores, features=None)[source]

Make records of feature subsets from flags and scores.

Parameters

  • flags ((M, N) list[list[bool]]) – Used features in subsets. Each row corresponds to a subset. Each column corresponds to a feature.

  • scores ((M,) list[float]) – Scores for feature subsets.

  • features ((N,) list[str] or None, optional) – Feature names. If None, features are named as feature_0, feature_1, etc.

Returns

pandas.Series – Score records.

Raises

ValueError – If the number of rows in flags is not equal to the size of scores. If the number of columns in flags is not equal to the size of features.

Notes

An index of returned pandas.Series is a pandas.MultiIndex indicating used features. Levels of the index are features.

Examples

>>> from_arrays([[True, False, False], [True, False, True]], [0.2, 0.8])
feature_0  feature_1  feature_2
True       False      False        0.2
                      True         0.8
dtype: float64

>>> from_arrays([[True, False, False], [True, False, True]], [0.2, 0.8],
...             features=['A', 'B', 'C'])
A     B      C
True  False  False    0.2
             True     0.8
dtype: float64