DataFrame does not work well with quantities

[tkf]

from pandas import DataFrame
import quantities as pq
df = DataFrame([[pq.Quantity(i, pq.s)] for i in range(3)])
df[1:] - df[:-1]

raises: ValueError: Unable to convert between units of "dimensionless" and "s"

Note that the following works w/o error because pandas "strips" dimension:

df = DataFrame(pq.Quantity(np.array([0,1,2]).T, pq.s))
df[1:] - df[:-1]

[changhiskhan]

normally, you'd be able to do something like:

nas = pq.Quantity(np.nan, pq.s)
df[1:].sub(df[:-1], fill_value=nas)

However, because pq.Quantity is an ndarray subclass there's some weird behavior. I'm not familiar with quantities internals, but here's what I saw just poking around:

In [54]: arr = np.array([np.nan, pq.Quantity(1, pq.s), pq.Quantity(2, pq.s)], dtype=object)

In [55]: isnull(arr)
Out[55]: array([ True, False, False], dtype=bool)

In [56]: arr[isnull(arr)] = pq.Quantity(np.nan, pq.s)

In [57]: arr
Out[57]: array([nan, 1 s, 2 s], dtype=object)

Essentially this makes the NaN filling code in pandas have not effect.

[tkf]

I noticed you can strip dimension by numpy.asarray. So I am using this function before giving any Quantity arrays to pandas.

[jreback]

closing in favor of #10349