Mailman 3 September 2018 - NumPy-Discussion

PR to add "weights" option to np.quantile
by Chun-Wei Yuan 13 Sep '18

13 Sep '18

Hi all, I have a long-standing PR to add a "weights" option to np.quantile/percentile: https://github.com/numpy/numpy/pull/9211 For a little background, there are quite a few ways to define "quantile" to begin with. Numpy defines it the same way as R's default (Type 7): https://www.rdocumentation.org/packages/stats/versions/3.5.0/topics/quantile What PR 9211 does is introducing a "weights" option while staying consistent with the Type 7 definition of quantile. There is certainly support on adding this feature, but as also can be perused from the thread, there are doubts on 1.) what is the right way to define "weighted" quantile? 2.) whether Numpy should support other types of quantiles in the first place, etc. I've stated my answers to those questions in the thread. This PR has fallen off the radar and needs a little popular jolt to get some movement. Please chime in to keep it alive. Best, Chun

1 0

ANN: PyData/Sparse 0.4.1
by Hameer Abbasi 12 Sep '18

12 Sep '18

============================= Announcing PyData/Sparse 0.4.1 ============================= (Apologies for the cross-posting) Hi everyone, This is a performance, bug-fix and feature release. The changelog can be seen at https://sparse.pydata.org/en/latest/changelog.html Highlights include: Faux In-place operations Mixed ndarray-sparse operations Fill-values other than zero Misc support for different functions What’s PyData/Sparse? ——————————— PyData/Sparse is a an N-dimensional sparse array library. It’s compatible with NumPy and follows the ndarray interface as closely as possible. It depends on NumPy, SciPy and Numba. Where can I find PyData/Sparse? ——————————————— The project is available on GitHub at https://github.com/pydata/sparse and is available on PyPI and conda-forge as “sparse”. Documentation is hosted as https://sparse.pydata.org. Best Regards, Hameer Abbasi

1 0

The http://planet.scipy.org/ site is down.
by Charles R Harris 12 Sep '18

12 Sep '18

Hi All, Anyone know who hosts http://planet.scipy.org/? Chuck

6 7

Re: [Numpy-discussion] A minor milestone
by Chris Barker 10 Sep '18

10 Sep '18

On Sat, Sep 8, 2018 at 9:17 PM, Charles R Harris <charlesr.harris(a)gmail.com> wrote: > There are probably a LOT of Windows users getting numpy from conda as well. >> >> I wonder if the conda folks have some statistics? > I'm sure there is a better way that I don't know, but anaconda.org does post number of downloads of pacakges: https://anaconda.org/search?q=+numpy So you've got conda-forge, and anaconda (which I think is the "default" channel for conda and Anaconda users) and even an Intel version -- that's a lot. But we have the CI problem -- conda-forge, for instance, uses CIs to build packages that depend on numpy -- so a LOT of those are downloaded to build other packages -- Im guessing that's why conda-forge has the most downloads of numpy (by a factor of 4) I suspect end users actually use defaults' numpy more. You can also see downloads by file for a given channel: https://anaconda.org/conda-forge/numpy/files?sort=ndownloads&sort_order=des… Which shows that py3 Linux is the most popular -- but again, probably CIs heavily influence that. Close-ish second is Linux py2.7, but 3.6 has 1.3 times as many -- so as the OP said -- py3 is seeing a lot of use! Even stronger for os-x I have no idea what conclusions to draw from any of this, but if someone can find an API to get this data, we could do more. And I think there's an issue with Windows and py2.7 -- it doesn't show up at all (probably the whole openblas / old compiler issue). So we dont know the stats for 2.7 vs 3.6 on Windows. -CHB > <snip> > > Chuck > > _______________________________________________ > NumPy-Discussion mailing list > NumPy-Discussion(a)python.org > https://mail.python.org/mailman/listinfo/numpy-discussion > > > -- Christopher Barker, Ph.D. Oceanographer Emergency Response Division NOAA/NOS/OR&R (206) 526-6959 voice 7600 Sand Point Way NE (206) 526-6329 fax Seattle, WA 98115 (206) 526-6317 main reception Chris.Barker(a)noaa.gov

2 1

A minor milestone
by Nathaniel Smith 10 Sep '18

10 Sep '18

Looking at https://pypistats.org/packages/numpy , it appears that August 24 was the last day when numpy had more Python 2 downloads than Python 3 downloads (maybe ever?). -n -- Nathaniel J. Smith -- https://vorpus.org

6 9

C99
by Charles R Harris 08 Sep '18

08 Sep '18

Hi All, I've a PR up converting travis testing to use C99 <https://github.com/numpy/numpy/pull/11905>. I suspect we may not want to merge it for a while, but it does raise a couple of style questions that we should probably settle up front. Namely: - Should we allow // style comments - Should we allow variable declarations after code I am sure there are others to consider that haven't occurred to me. I confess that I am not a big fan of allowing either, but am probably prejudiced by early familiarity with C89 and long years working to that spec. Thoughts? Chuck

3 5

scaling of the covariance matrix in np.polyfit
by Andreas Nußbaumer 06 Sep '18

06 Sep '18

Hi, as previously discussed on the mailing list http://numpy-discussion.10968.n7.nabble.com/Inconsistent-results-for-the-co… and listed in the issude https://github.com/numpy/numpy/issues/11196 there is no-standard factor to the scaling of the covariance matrix in np.polyfit. A while ago, I posted the pull request https://github.com/numpy/numpy/pull/11197 that removes the non-standard factor and introduces an option to give out the unscaled covariance. After an inital review by Marten van Kerkwijk, the only problem left was the question about the api. I have not heard back in while and was wondering whether there is still interest in this PR. Greetings Andreas

1 0

ARMv8 / shippable addition to CI
by Tyler Reddy 06 Sep '18

06 Sep '18

Hi, Stefan & Matti wanted me to check on the mailing list about about adding ARMv8 (free / shippable service) to NumPy CI on github -- is anyone opposed to that? Only obvious drawback seems to be that the service sometimes seems a little unreliable, not that this is new to us; guess we could just turn it off if so. The issue with progress: https://github.com/numpy/numpy/issues/11702 All NumPy tests pass (regular not --full suite anyway) on Python 2.7 / 3.7 in 9 minutes if we cache the Cython build: https://app.shippable.com/github/tylerjereddy/numpy/runs/56/summary/console If we move forward: I've been testing on a personal branch so that I didn't have to act on behalf of NumPy officially--does someone on the steering council want to set up an official NumPy shippable account & ask them for access (I think I had to push a button / send an email for ARMv8 specifically). I can open a PR to NumPy proper if this is generally +1 now. Tyler

3 2

Re: [Numpy-discussion] Adoption of a Code of Conduct
by Peter Creasey 01 Sep '18

01 Sep '18

+1 for keeping the same CoC as Scipy, making a new thing just seems a bigger surface area to maintain. Personally I already assumed Scipy's "honour[ing] diversity in..." did not imply any protection of behaviours that violate the CoC *itself*, but if you wanted to be really explicit you could add "to the extent that these do not conflict with this code of conduct." to that line. Best, Peter

8 13