Re: [Numpy-discussion] Cross-correlation PR stuck in limbo

[Jaime Fernández del Río]

I did an overall review of the code a couple of weeks ago (see the PR for
details), and there is quite some work to be done before we can merge
Honi's code. But if he can find the time to work on the coding, I'll try to
be more diligent about the reviewing.

Jaime

On Fri, May 27, 2016 at 12:51 PM, Elliot Hallmark 
wrote:

> +1
>
> This would really help with large data sets in certain situations.
>
> Is there still disagreement about whether this should be included? Or are
> there some minor details still? Or just lost in the shuffle?
>
> Hopefully,
>   Elliot
>
> On Wed, May 4, 2016 at 7:07 AM, Pierre Haessig 
> wrote:
>
>> Hi,
>>
>> I don't know how to push the PR forward, but all I can say is that this
>> maxlag feature would be a major improvement for using Numpy in time
>> series analysis! Immediate benefits downstream for Matplotlib and
>> statsmodel.
>>
>> Thanks Honi for having taken the time to implement this!
>>
>> best,
>> Pierre
>>
>> ___
>> NumPy-Discussion mailing list
>> NumPy-Discussion@scipy.org
>> https://mail.scipy.org/mailman/listinfo/numpy-discussion
>>
>
>
> ___
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> https://mail.scipy.org/mailman/listinfo/numpy-discussion
>
>


-- 
(\__/)
( O.o)
( > <) Este es Conejo. Copia a Conejo en tu firma y ayúdale en sus planes
de dominación mundial.
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
https://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: [Numpy-discussion] Cross-correlation PR stuck in limbo

[Elliot Hallmark]

+1

This would really help with large data sets in certain situations.

Is there still disagreement about whether this should be included? Or are
there some minor details still? Or just lost in the shuffle?

Hopefully,
  Elliot

On Wed, May 4, 2016 at 7:07 AM, Pierre Haessig 
wrote:

> Hi,
>
> I don't know how to push the PR forward, but all I can say is that this
> maxlag feature would be a major improvement for using Numpy in time
> series analysis! Immediate benefits downstream for Matplotlib and
> statsmodel.
>
> Thanks Honi for having taken the time to implement this!
>
> best,
> Pierre
>
> ___
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> https://mail.scipy.org/mailman/listinfo/numpy-discussion
>
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
https://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: [Numpy-discussion] Cross-correlation PR stuck in limbo

[Pierre Haessig]

Hi,

I don't know how to push the PR forward, but all I can say is that this
maxlag feature would be a major improvement for using Numpy in time
series analysis! Immediate benefits downstream for Matplotlib and
statsmodel.

Thanks Honi for having taken the time to implement this!

best,
Pierre

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
https://mail.scipy.org/mailman/listinfo/numpy-discussion

[Numpy-discussion] Cross-correlation PR stuck in limbo

[Honi Sanders]

Hello all,
I have completed a pull request to add a “maxlag” functionality to 
numpy.correlate.  See here: https://github.com/numpy/numpy/pull/5978 
.  This pull request has passed all 
tests and has been ready to be merged for around six months.  Several people 
have commented requesting for it to be included on stackoverflow, the 
listserve, and github.  Can someone please let me know what needs to be done or 
can it be merged?

Here is some background:
What was troubling me is that numpy.correlate does not have a maxlag feature. 
This means that even if I only want to see correlations between two time series 
with lags between -100 and +100 ms, for example, it will still calculate the 
correlation for every lag between -2 and +2 ms (which is the length of 
the time series). This (theoretically) gives a 200x performance hit! 

I have introduced this question as a numpy issue 
, a scipy issue 
 and on the scipy-dev list 
. It seems the 
best place to start is with numpy.correlate, so that is what I am requesting. 

Previous discussion of this functionality can be found at another discussion on 
numpy correlate (and convolution) 
.
 Other issues related to correlate functions include ENH: Fold fftconvolve into 
convolve/correlate functions as a parameter #2651 
, Use FFT in np.correlate/convolve? 
(Trac #1260) #1858 , and normalized 
cross-correlation (Trac #1714) #2310 
.



The new implementation allows new types of the “mode” argument, to include an 
int value, which defines the maximum lag for which cross-correlation should be 
calculated, or a tuple, which defines the minlag, maxlag, and lagstep to be 
used in the same format as the arguments to numpy.arange.


Please let me know what should be done to move this pull request forward.

Honi



___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
https://mail.scipy.org/mailman/listinfo/numpy-discussion