Pandas - select values from a groupby object

Question

I have a pandas groupby object, c:

>>> c.index.names
FrozenList([u'Thing1', u'Thing2', u'Month'])
>>> c.columns
Index([u'Tot'], dtype='object')
>>> c
                           Tot
Thing1 Thing2 Month                
G      P      2012-12-01   0.017640
              2013-01-01   0.012062
              2013-02-01   0.029022
              2013-03-01   0.007593
              2013-04-01   0.004862
              2013-05-01   0.002671
              2013-06-01   0.014895
              2013-07-01   0.029641
              2013-08-01   0.051129
              2013-09-01   0.023913
              2013-10-01   0.061406
              2013-11-01   0.054781
              2014-01-01   0.017115
              2014-02-01   0.011919
H      K      2013-06-01   2.390632
              2013-07-01   7.066034
              2013-08-01   5.426312
              2013-09-01   8.276066
              2013-10-01   5.745811
              2013-11-01   2.250162
              2013-12-01   0.976822
              2014-01-01   1.438316
              2014-02-01   3.507220
       M      2012-06-01   3.050136
              2012-07-01   5.911788
              2012-08-01   2.794381
              2012-09-01   4.418268
              2012-10-01   5.312635
              2012-11-01   1.810977
              2012-12-01   3.097878
              2013-01-01   0.811326
              2013-02-01   3.105154
              2013-03-01   2.384704

How do you select the timeseries values for Thing1=G and Thing2=P, for plotting?

Answer 1

You can just index it like other dataframes, but using tuples for the combination of levels:

c.loc[('G', 'P'), :]

However, it is possible that for such indexing operations, the multi-index needs to be sorted (otherwise it can give and exception). You can do this with:

c = c.sortlevel()

---

And to be correct, c is not a groupby object, but a DataFrame (you also have pandas GroupBy objects, but they are the result of a .groupby() call)