Meeting notes 2020 October 26th

meeting_notes/2020-10-26.md

@@ -0,0 +1,50 @@
+# October 26th 2020
+
+### General topics
+- [0.24 Release](https://github.com/scikit-learn/scikit-learn/pulls?q=is%3Aopen+is%3Apr+milestone%3A0.24):
+  - pair releasing: Olivier and Chiara
+  - Chiara will review the [milestone](https://github.com/scikit-learn/scikit-learn/pulls?q=is%3Aopen+is%3Apr+milestone%3A0.24) list, remove items if necessary,
+    ping reviewers or contributors if necessary
+  - pull requests from the milestone will be pushed during this week, next Monday Olivier will create the release branch
+  - Olivier: HGBT and categorical features is still challenging; we don't need to block the PR and can handle the ease of use in future PRs.
+  - Check CircleCI checks: nice to fix all the failures for the release.
+  -  documentation and nightly builds are no longer hosted on rackspace, but at anaconda.org ([#18652](https://github.com/scikit-learn/scikit-learn/pull/18652))
+
+### Need decision
+- [SLEP015](https://github.com/scikit-learn/enhancement_proposals/pull/48): **priority for 1.0**, will be discussed during the next meeting with Andy and Thomas
+
+### Need attention (review)
+(Christian)
+- [17386](https://github.com/scikit-learn/scikit-learn/pull/17386) Poisson Decision Trees has 2 approvals, but a good look before merging would be nice.
+- [17027](https://github.com/scikit-learn/scikit-learn/issues/17027) SplineTransformer, but not for the coming release.
+- [15123](https://github.com/scikit-learn/scikit-learn/issues/15123) Common Private Loss Functions, technical problems seem solved. Draft PR almost ready.
+  How to proceed? What are criteria for inclusion? Second opinion would help a lot with design.
+  - Find a compromise: identify some models that could use the same loss functions. Do not try to homogenize all the modules
+  - Use tests to check that different implementations give same results
+  - liblinear and libsvm are out of scope
+
+(Guillaume)
+ - [18543](https://github.com/scikit-learn/scikit-learn/pull/18543) First PR linked to plotting API.
+ - [18589](https://github.com/scikit-learn/scikit-learn/pull/18589) Refactoring to create a `_get_response` to get prediction from one of `predict`, `predict_proba` or `decision_function`.
+   - When this is reviewed, make sure that we do not overlook the complexity of the multi-label cases
+   - Related (but not a block for the above) unify the shape of the output of these functions?
+       - related: **https://github.com/scikit-learn/scikit-learn/issues/9650**
+       - Will have to bite this bullet before 1.0
+
+(Adrin)
+  - [SLEP06](https://scikit-learn-enhancement-proposals.readthedocs.io/en/latest/slep006/proposal.html) and
+    [related prototype](https://github.com/scikit-learn/scikit-learn/pull/16079):
+    - discuss in a dedicated call: how to move for a vote and choose a prototype 
+    - Adrin will ask for people who want to be on the call and organize that
+
+- Nicolas asking for a different process for SLEP review with more opportunities to raise questions and concerns.
+
+### Contributors
+
+
+### Next meeting
+Two options:
+- https://www.timeanddate.com/worldclock/meetingdetails.html?year=2020&month=11&day=30&hour=16&min=0&sec=0&p1=179&p2=240&p3=195&p4=224
+- https://www.timeanddate.com/worldclock/meetingdetails.html?year=2020&month=11&day=30&hour=20&min=0&sec=0&p1=179&p2=240&p3=195&p4=224
+
+Chiara will send a doodle to select one of those two options.