Add scikit-learn

[GaelVaroquaux]

We'd like to request enrolling scikit-learn into LFX Insights. The GitHub org is https://github.com/scikit-learn/scikit-learn.

Scikit-learn is the most used machine learning library: download numbers can be compared across projects https://clickpy.clickhouse.com/dashboard/scikit-learn

[philipjk]

is the number of downloads from PyPI a good proxy metrics for usage in ML? even if downloads were a good proxy metrics for usage, lots of people in ML use conda (especially for torch/tf with GPU), whose downloads are not tracked there

[chrisvoncsefalvay]

If there has ever been a package that did not necessarily need download numbers to attest to its ubiquity in this field, it’s sklearn. Not only is it one of the most widely cited python packages, a mainstay of research in academia and commercial applications alike and a de facto industry standard, it’s also a source of primitives for lots of other packages I don’t disagree download numbers alone do not tell the whole story. Rather, I’d suggest that sklearn ‘s download numbers are only a nice quantification of its massive impact.

[GaelVaroquaux]

Download numbers are always difficult to interpret, but they are indeed a proxy for usage (some of it is usage by bots, or compute cluster, rather than human beings, but it is still usage).

Regarding your question on conda, I did an analysis across all conda channels (thus including the pytorch specific channel), and the results are similar:

Analysis can be found here: https://colab.research.google.com/drive/1sMzVVPm-9OO42ZugBCqbWmHOekfzn-hV?usp=sharing

Moving beyond download numbers, another way to look at usage is to look at the "used by" entry visible on github (on the lower right of the project pages):

• For scikit-learn

• For pytorch

These numbers are easy to check for any project on github

[jonathimer]

I started the onboarding. The project should be available on Insights within the next 48 hours.

[GaelVaroquaux]

It is indeed on the Insights.

Thank you so much, you really made my day.

[jonathimer]

You're very welcome!