DOC Release highlights for version 0.24 #18795
Conversation
There was a problem hiding this comment.
Thanks @cmarmo, I commented below where I noted a few inaccuracies.
SH is really hard to explain, let alone in such a short amount of text. Maybe the safest way here is to copy-paste the first 2 paragraphs of the UG, with some minor subsampling / alterations?
|
New examples are usually not listed in release notes. Could we advertise those a bit more, somewhere? |
|
Maybe add |
In principle release highlights are linked in the homepage |
I think the point of @lorentzenchr is to link some important, newly introduced examples from the 0.24 release highlight example itself. For instance: https://scikit-learn.org/dev/auto_examples/model_selection/plot_grid_search_stats.html |
Sorry I misunderstood... I think a nice solution would be to have a "New" tag like suggested in #16601 |
Yes, but it would also be worth adding a paragraph /section in the release highlights that lists new or significantly improved, well-developed examples (with a short description and a link). |
|
@lorentzenchr @ogrisel, I have added a section about documentation improvements using also html meta tags ... this is a suggestion for #16601 also. |
|
I'm wondering if we can add in some way the information about the new build system via github actions directly in the scikit-learn repo. |
I don't think so. It's not a new feature for the users. The real impact is for the maintainers. |
|
I'm wondering if we can add in some way the information about the new build system via github actions directly in the scikit-learn repo.
I'm not sure that this is something that the end user will get excited about.
|
Ok. Let's keep it simple then. |
There was a problem hiding this comment.
Thanks a lot @cmarmo, probably last comments from me, LGTM when addressed.
The current highlights run in about 48 seconds, which is quite long compared to the previous ones. Any chance we can reduce the timing a bit? Perhaps we can use LogisticRegression instead of KNeighborsClassifier with SFS? Maybe the SH and ICE parts are slow too, I haven't run it.
Co-authored-by: Nicolas Hug <contact@nicolas-hug.com>
Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>
|
@NicolasHug feel free to finalize this PR as you prefer: this will move things faster wrt some new back and forth about my copied and pasted text. |
|
I've pushed my remaining comments. I'll take a last look at the rendered html and will merge when green. |
|
do we have some broken stuff here |
|
Oh no, just some git checkout errors due merging in master before the build starts :) |
|
Should be fine, I merged without waiting for all checks to finish because I only made minor doc changes in the last commits |
|
Thanks @Chiara and @NicolasHug! |
What does this implement/fix? Explain your changes.
Adds release highlights for version 0.24
Any other comments?
This is still very WIP, but I have a list of features I'm planning to highlight: @scikit-learn/core-devs do you mind letting me know if there are some missing or unnecessary? Thanks a lot!
@NicolasHug, successive halving is drafted already, let me know if I'm on the right direction here. Thanks!