8000 BUG: Added proper handling of median and percentile when nan's are prese... by empeeu · Pull Request #5753 · numpy/numpy · GitHub
[go: up one dir, main page]

Skip to content

BUG: Added proper handling of median and percentile when nan's are prese... #5753

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jun 23, 2015

Conversation

empeeu
Copy link
Contributor
@empeeu empeeu commented Apr 8, 2015

...nt in array to close issue #586.

Also added unit tests.

@charris
Copy link
Member
charris commented Apr 9, 2015

@juliantaylor Want to have a look?

@charris
Copy link
Member
charris commented Jun 3, 2015

@juliantaylor ping.

@homu
Copy link
Contributor
homu commented Jun 17, 2015

☔ The latest upstream changes (presumably #5977) made this pull request unmergeable. Please resolve the merge conflicts.

@empeeu
Copy link
Contributor Author
empeeu commented Jun 23, 2015

@homu I rebased again. Should be able to merge now.

@charris
Copy link
Member
charris commented Jun 23, 2015

@empeeu homu is a github bot ;)

@empeeu
Copy link
Contributor Author
empeeu commented Jun 23, 2015

O rly? That's good to know. Is it new?

@njsmith
Copy link
Member
njsmith commented Jun 23, 2015

Pretty new:
http://homu.io/
http://mail.scipy.org/pipermail/numpy-discussion/2015-June/073058.html

On Mon, Jun 22, 2015 at 6:03 PM, Ueckermann notifications@github.com
wrote:

O rly? That's good to know. Is it new?


Reply to this email directly or view it on GitHub
#5753 (comment).

Nathaniel J. Smith -- http://vorpus.org

@empeeu
Copy link
Contributor Author
empeeu commented Jun 23, 2015

That's pretty cool. Never break again!

@juliantaylor
Copy link
Contributor

I'm still not very happy about this change, but its probably more my irrational longing for optimal performance on well behaved data. As most people were for it when it was discussed, in it goes.
Thanks for the contribution and the patience.

juliantaylor added a commit that referenced this pull request Jun 23, 2015
BUG: Added proper handling of median and percentile when nan's are prese...
@juliantaylor juliantaylor merged commit e4ab830 into numpy:master Jun 23, 2015
@madphysicist
Copy link
Contributor

As part of creating an iqr function in scipy.stats, I am relying on this functionality being present in numpy versions after 1.11.0b3. Is there any reason, (such as backports or something else) why my assumption is not correct?

@juliantaylor
Copy link
Contributor

it is unlikely something like this would get backported as its quite invasive so its probably fine, but in general you should always test for behaviour instead of version numbers, it is normal for distributions to backport patches.

@njsmith
Copy link
Member
njsmith commented Feb 16, 2016

Also it's generally nice to wait until an actual release (like 1.11.0) before counting on anything, because the whole point of a prerelease cycle is that we sometimes discover problems and have to revert things. Maintaining backcompat is hard enough to start with without also having to satisfy downstr A80D eam assumptions about behavior from the future :-)

@madphysicist
Copy link
Contributor

@juliantaylor and @njsmith. Thanks for the advice. My problem is that there have been quite a few changes to percentile and nanpercentile starting with version 1.9.0, so I have a complex decision tree that mostly throws warnings about incomplete functionality. I will do my best to test for behavior where possible, but in the worst case I will just use actual release numbers. Hopefully the scipy maintainers are a little more forgiving of try...except :-)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants
0