A B
DATE
2013-05-01 473077 71333
2013-05-02 35131 62441
2013-05-03 727 27381
2013-05-04
Simply using df.mean()
will Do The Right Thing(tm) with respect to NaNs:
>>> df
A B
DATE
2013-05-01 473077 71333
2013-05-02 35131 62441
2013-05-03 727 27381
2013-05-04 481 1206
2013-05-05 226 1733
2013-05-06 NaN 4064
2013-05-07 NaN 41151
2013-05-08 NaN 8144
2013-05-09 NaN 23
2013-05-10 NaN 10
>>> df.mean(axis=1)
DATE
2013-05-01 272205.0
2013-05-02 48786.0
2013-05-03 14054.0
2013-05-04 843.5
2013-05-05 979.5
2013-05-06 4064.0
2013-05-07 41151.0
2013-05-08 8144.0
2013-05-09 23.0
2013-05-10 10.0
dtype: float64
You can use df[["A", "B"]].mean(axis=1)
if there are other columns to ignore.