Behavior of numeric ops on Index has changed

[bryevdv]

The following used to work:

In [7]: import pandas as pd

In [8]: import numpy as np

In [9]: a = np.zeros(5, dtype=np.float)

In [10]: df = pd.DataFrame({"foo": [10, 20, 30, 40, 50]})

In [11]: df.index
Out[11]: Int64Index([0, 1, 2, 3, 4], dtype='int64')

In [12]: df.index * 3.2
Out[12]: Int64Index([0.0, 3.2, 6.4, 9.6, 12.8], dtype='float64')

In [13]: a - df.index * 3.2
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-13-8a6574b7beca> in <module>()
----> 1 a - df.index * 3.2

TypeError: unsupported operand type(s) for -: 'numpy.ndarray' and 'Int64Index'

In [14]: pd.__version__
Out[14]: '0.15.0'

[jreback]

@bryevdv workaround till fixed is this:

In [19]: a-df.index.to_series()*3.2
Out[19]: 
0     0.0
1    -3.2
2    -6.4
3    -9.6
4   -12.8
dtype: float64

[TomAugspurger]

also

In [23]: np.subtract(a, df.index) * 3.2
Out[23]: Int64Index([0.0, -3.2, -6.4, -9.6, -12.8], dtype='float64')

[cpcloud]

Also strange that it's returning an integer index with float values.

[bryevdv]

Yes that's why I included that line but I should have mentioned explicitly.

On Oct 23, 2014, at 07:17, Phillip Cloud [email protected] wrote:

Also strange that it's returning an integer index with float values.

—
Reply to this email directly or view it on GitHub.