[Numpy-discussion] Computing Simple Statistics When Only they Frequency Distribution is Known

Anne Archibald peridot.faceted@gmail....
Sun Nov 29 19:15:46 CST 2009


2009/11/28 Wayne Watson <sierra_mtnview@sbcglobal.net>:
> Anne Archibald wrote:
>> 2009/11/28 Wayne Watson <sierra_mtnview@sbcglobal.net>:
>>
>>> I was only illustrating a way that I would not consider, since the
>>> hardware has already created the pdf. I've already coded it pretty much
>>> as you have suggested. As I think I mention ed above, I'm a bit
>>> surprised numpy doesn't provide the code you suggest as part of some
>>> function. CalcSimplefromPDF(xvalues=mydatarray, avg=ture, minmax=true,
>>> ...).
>>>
>>
>> Feel free to submit an implementation to numpy's issue tracker. I
>> suggest modifying mean, std, and var (at least) so that, like average,
>> they take an array of weights.
>
> How would I do that?
>

Obtain a copy of recent numpy source code - a .tar file from the
website, or using SVN, or git. Then add the feature plus some tests,
confirm that the tests pass, and post a request and your patch to the
bug tracker.

Anne


More information about the NumPy-Discussion mailing list