In [16]:
each_period=cohorts['FirstOrder'].value_counts().sort_index()
each_period
Out[16]:
2009-01    15
2009-02    13
2009-03    13
2009-04    12
2009-05    11
2009-06    10
2009-07     9
2009-08     8
2009-09     7
2009-10     6
2009-11     5
2009-12     4
2010-01     3
2010-02     2
2010-03     1
Name: FirstOrder, dtype: int64