error in row indexing using loc

[jkokorian]

I am trying to get the value of a dataframe column that contains numpy arrays. The index is a timestamp.The dataframe column has the dtype 'O', which is essential for reproducing the bug.

import pandas as pd
import numpy as np
from pandas import Timestamp

timestamps = pd.Series(data=[Timestamp('2014-07-01 11:19:07.888000', tz=None),Timestamp('2014-07-01 11:19:08.094000', tz=None),Timestamp('2014-07-01 11:19:08.299000', tz=None)],name='timestamp')
data = pd.Series(dtype='O',data=[np.array([1,2,3]),np.array([4,5,6]),np.array([7,8,9])],index=timestamps)
df = pd.DataFrame(data=data,columns=['a'])

i = df.index[0]

#this works as intended, produces a single row of the dataframe
df.loc[i]

#this gives an error (which it shouldn't do). It should have produced the array [1,2,3]
df.a.loc[i0]

#this works as a workaround, but it's a bit ugly
df.a.get_value(i)

I really hopes that this can get fixed, thanks!

[jreback]

already closed by #6396

keep in mind that this is a really really odd thing to do
keeling an embedded array in a series (or frame)
you lose all efficiency of numpy/pandas

much much much better is to use a data frame in this case (or a panel)
if you need higher dimensionaility