[
https://issues.apache.org/jira/browse/ARROW-9056?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17127710#comment-17127710
]
Wes McKinney commented on ARROW-9056:
-------------------------------------
Seems reasonable. For NumPy it's a no-op except in the case where a minimum
numbers of observations is required (e.g. standard deviation with ddof > 0)
{code}
In [1]: import numpy as np
In [2]: np.sum(5)
Out[2]: 5
In [3]: np.mean(5)
Out[3]: 5.0
In [4]: np.std(5)
Out[4]: 0.0
In [5]: np.std(5, ddof=1)
C:\Miniconda\envs\pyarrow-dev\lib\site-packages\numpy\core\_methods.py:217:
RuntimeWarning: Degrees of freedom <= 0 for slice
keepdims=keepdims)
C:\Miniconda\envs\pyarrow-dev\lib\site-packages\numpy\core\_methods.py:209:
RuntimeWarning: invalid value encountered in double_scalars
ret = ret.dtype.type(ret / rcount)
Out[5]: nan
{code}
so (sample) standard deviation would probably return null but the others would
just return the scalar unmodifed
> [C++] Aggregation methods for Scalars?
> --------------------------------------
>
> Key: ARROW-9056
> URL: https://issues.apache.org/jira/browse/ARROW-9056
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Neal Richardson
> Priority: Major
> Fix For: 1.0.0
>
>
> See discussion on https://github.com/apache/arrow/pull/7308. Many/most would
> no-op (sum, mean, min, max), but maybe they should exist and not error? Maybe
> they're not needed, but I could see how you might invoke a function on the
> result of a previous aggregation or something.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
