Sampling From Group

[jeff-hernandez]

Closes #124 by adding the parameter per_instance, which determines whether or not the sampling should be applied to each group. Working through the code, it made sense to make this parameter boolean. This added more functionality by allowing users to sample a fraction of the labels from each group.

Example

>>> lt

   entity labels
0       0      a
1       0      a
2       0      b
3       0      b
4       1      a
5       1      a
6       1      b
7       1      b

>>> frac = {'a': 1, 'b': .5}
>>> lt.sample(frac=frac, per_instance=True)

   entity labels
0       0      a
1       0      a
2       0      b
4       1      a
5       1      a
7       1      b

[codecov]

Codecov Report

Merging #144 into label_select will not change coverage.
The diff coverage is 100.00%.

@@              Coverage Diff               @@
##           label_select      #144   +/-   ##
==============================================
  Coverage        100.00%   100.00%           
==============================================
  Files                26        26           
  Lines              1048      1069   +21     
==============================================
+ Hits               1048      1069   +21     

Impacted Files	Coverage Δ
composeml/label_times/object.py	100.00% <100.00%> (ø)
...mposeml/tests/test_label_transforms/test_sample.py	100.00% <100.00%> (ø)

[frances-h]

LGTM 👍