scikit-learn / communication Goto Github PK
View Code? Open in Web Editor NEWIssues and content related to communicating about scikit-learn
License: BSD 3-Clause "New" or "Revised" License
Issues and content related to communicating about scikit-learn
License: BSD 3-Clause "New" or "Revised" License
Ask Twitter if the scikit-learn account can be a verified one
Now that we have 100+ subscribers on YouTube, we can set up a custom URL:
https://www.youtube.com/channel/UCJosFjYm0ZYVUARxuOZqnnw
cc: @laurburke
Here, please include what you would like to share. If you have suggested text, feel free to include as well.
cc: @laurburke
Given the recent news and the philosophy of Mastodon. We discussed the possibility of creating a Mastodon account for scikit-learn with @ogrisel. What do you think about it?
scikit-learn/scikit-learn#22868 and it's enabling PR scikit-learn/scikit-learn#22760 could be worth a post.
See Arliss email dated 2022-02-11 with subject "[CDR] Outreach for Interview Participants"
Could you please reach out to your contributor community to ask if anyone is interested in participating in an interview. Here is the “Participant Interest” form to capture people’s contact information.
I’ve attached a couple of documents to help you with this …
Sample text that you can either copy and paste directly or adapt to best suit your contributor audience.
A copy of the questions that are included in the “Participant Interest” form.
Feel free to use our text or craft something of your own, and include the form link**
For information, Thomas started to draft some release highlights for 1.1 in scikit-learn/scikit-learn#23084.
Here is the HTML rendering of the current state of the PR : https://output.circle-artifacts.com/output/job/d83a2974-e9d4-479b-814d-46bb0fa99489/artifacts/0/doc/auto_examples/release_highlights/plot_release_highlights_1_1_0.html
I recently made a speech at the PyLadies meetup.
There is a live video there: https://www.youtube.com/watch?v=GfkSvUug7js&ab_channel=PyladiesParis
The link to the slides is the following: https://docs.google.com/presentation/d/14N8f3wGDG25Yz6mBXDvpsnbzTII_RYR7WsFEXh10H94/edit?usp=sharing
Feel free to share it (or close the issue) if you think it worth sharing.
scikit-learn swag options are up, added by NumFOCUS:
https://numfocus.myspreadshop.com/all
https://numfocus.myspreadshop.com/scikit-learn+logo?idea=6335cad48f3f5268f5f42559
For twitter:
Did you know you can do NLP with scikit-learn? Learn the basics of text vectorizers with this mini tutorial (1/3)
https://scikit-learn.org/dev/auto_examples/text/plot_hashing_vs_dict_vectorizer.html
Learn the basics on classification of text documents with this mini tutorial (2/3)
https://scikit-learn.org/dev/auto_examples/text/plot_document_classification_20newsgroups.html
Learn the basics of text documents clustering with this mini tutorial (3/3)
https://scikit-learn.org/dev/auto_examples/text/plot_document_clustering.html
I am biased because I am a co-author but it would be nice that scikit-learn relays this scikit-learn blog post: https://blog.scikit-learn.org/updates/community/joining-forces-hugging-face/
and the Hugging Face social media posts:
https://twitter.com/huggingface/status/1582029499775930370
https://www.linkedin.com/feed/update/urn:li:activity:6987796152913911808/
Cheers
Perhaps we are in a place now where we can "claim" the old FB page and merge the followers from old page into ours?
Current FB page (owned by scikit-learn):
https://www.facebook.com/scikitlearnofficial
Older FB page: 871 followers
https://www.facebook.com/ScikitLearn
Resource: claim FB page
https://www.reviewtrackers.com/blog/claim-facebook-business-page/
cc: @laurburke
As of now, we have 8000+ followers on LinkedIn:
https://www.linkedin.com/company/scikit-learn/
It would be nice to do a celebratory post when we hit 10 000 followers
Example:
https://www.linkedin.com/in/data-umbrella-b7a64a1b1/
Good practice to connect Buffer account with an admin LI account, rather than a personal LI account.
cc: @laurburke
While our audience is familiar with the scikit-learn package, it'd be great to help them put faces to the names of those who make it all possible!
I'll be creating a short survey and a fun Gif template for us to promote on social. Participation is optional, but encouraged for well-deserved recognition!
We could probably improve the scikit-learn wikipedia pages in terms of update and structure/content.
Why?
Action: Having in mind the different persona of readers, I think we could also come up with a new plan and structure with high to low level of information granularity.
How to communicate about scikit-learn.
Will include information on:
announced on the public mailing list: https://mail.python.org/pipermail/scikit-learn/2022-September/004295.html
2022-09-19:
The core developers for Scikit-learn have recently voted to welcome Meekail
Zain to the team, recognizing Zain's efforts and trustworthiness as a
contributor. Congratulations, and thank you!
Share sprints blog on social media:
https://blog.scikit-learn.org/events/sprints-value/
A series of images/gifs we can use to create a cohesive social representation of the scikit-learn brand.
I have added a very pretty basic page to give a warning for people visiting our ".com" domain, rather than the ".org" domain (purposely not linking here, to avoid increasing the page rank of this page):
https://github.com/scikit-learn/scikit-learn.com
I used a very basic and empty hugo theme. My gut feeling is that we want very little here because we do not want people to link here. For this reason, I did not do a redirect in any way.
The number one purpose of owning this domain name is to avoid typosquatting on it.
I think that the comm's team should have a quick look to decide whether my work is good enough (in which case: close this issue), or whether it should be improved (in which case: work on the corresponding repo)
I am going to make a tutorial at the upcoming PyData Berlin: https://2022.pycon.de/program/9LZTRR/
I can provide any further information if required (I still need to prepare the talk thought :))
Here is a list of international days and weeks that can be good opportunities to convey related and specific messages to a broader audience using dedicated #.
The structure of this issue is:
Month
April
May
June
October
November
December
We will soon reach 50,000 github stars. It would be nice to communicate around that.
The PRs scikit-learn/scikit-learn#21808, scikit-learn/scikit-learn#22548 and, source of the inmprovements, scikit-learn/scikit-learn#20567 could be worth a short post.
@jeremiedbb noticed that we will hit a record of both opened and closed PRs in March (source)
We still need to wait for the very end figures but would be nice to schedule posts about this outstanding march summary.
I think scikit-learn/scikit-learn#21800 is worth mentioning. scikit-learn/scikit-learn#20567 was the enabler PR for it.
In the flavor of SLEP 19 aiming to recognize different types of contribution beyond code, @jjerphan and I suggest we add an allcontributors bot section to the readme or maybe elsewhere not to overload the scikit-learn readme.
A running example of this is the readme of Array API
What are your thoughts about this?
mix_params = {
'onehot__max_categories': [3, 6, 9, 12],
'rf__n_estimators': [10, 20],
'rf__max_depth': [2, 5],
'rf__min_samples_split': [2, 5],
'rf__min_samples_leaf': [2, 5],
'rf__max_features': ['auto', 'sqrt', 'log2', None],
'rf__bootstrap': [True, False]
}
pipe = Pipeline([
('onehot', OneHotEncoder(sparse_output=False)),
('rf', RandomForestClassifier(random_state=42))
])
model = GridSearchCV(pipe,
param_grid=mix_params,
scoring=['roc_auc', 'accuracy', 'recall', 'precision'],
refit='accuracy',
return_train_score=True,
cv=5, n_jobs=1)
model.fit(X, y)
model.cv_results_['mean_test_accuracy']
In this case, a Pipeline is used in GridSearchCV. but model.best_score_
and model.best_index_
are nan and 0, which means all testing scores are NaN, is anyone know how is this happening and how to fix?
For twitter:
Next Office Hour this Monday!
The focus of this session is to answer questions about contributing to scikit-learn.
When: 10:00 PT / 13:00 ET / 18:00 UTC / 19:00 CET
Where: https://buff.ly/3sTK2vD
Calendar: https://buff.ly/3yXGMmK
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.