Apply fails on empty Series

[alansaul]

I would expect the sum of [] to be giving me 0 as numpy would:

In [18]: import numpy as np

In [19]: np.sum([])
Out[19]: 0.0

but it appears apply cannot handle a lambda with a sum within if its logical selection evaluates to a empty series.

In [15]: a = pd.Series([])

In [16]: a.apply(lambda x: sum(a == x))
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-16-a2ddbb49d4e4> in <module>()
----> 1 a.apply(lambda x: sum(a == x))

/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-    packages/pandas/core/series.pyc in apply(self, func, convert_dtype, args, **kwds)
   2295 
   2296         mapped = lib.map_infer(values, f, convert=convert_dtype)
-> 2297         if isinstance(mapped[0], Series):
   2298             from pandas.core.frame import DataFrame
   2299             return DataFrame(mapped.tolist(), index=self.index)

IndexError: index 0 is out of bounds for axis 0 with size 0

[hayd]

This returns Series([], dtype: float64) on the latest pandas, which version of pandas are you using?

I think @changhiskhan fixed this in #2476 (0.10) ? :)

[ghost]

if the apply func argument is not a ufunc, apply acts like map with a lambda - the result is a series of values,
for an empty series you get an empty series of results.

The exception you show has been fixed.

closing as can't repro.