Request trove classifier for scikit-learn

[fkromer]

Request to add a new Trove classifier.

The name of the classifier you would like to add

• scipy :: scikit-learn
• scipy :: learn

Why do you want to add this classifier?

scikit-learn is one of the most popular "general purpose" machine learning frameworks for Python. There are many Python packages extending the functionality of scikit-learn, refer to this incomplete list of PyPi packages (section algorithms and extensions). A search for "sklearn" on pypi.org has 441 hits right now. scikit-learn is an extension of SciPy, one of the most popular Python ecosystems for mathematics, science, and engineering. SciPy extensions are called scikits. A search for "scikit" on pypi.org has 1130 hits right now. A trove classifier would help people to find suitable packages on pypi and would allow to process package meta data automatically using the Warehouse APIs. One of the proposed classifiers above could be suitable for scikit-learn.

[jamadden]

Hi! Thanks for the well written request. I suspect the classifiers being requested need to be in the Framework namespace, i.e., Framework :: SciPy :: scikit-learn. The PyPI moderators are in the process of developing a few guidelines for the creation of new Framework classifiers. Because removing classifiers is very difficult, and the longer the classifier list gets the harder it is to browse, I think we're expecting the bar to add new ones to be relatively high. Currently, we're looking at these guidelines:

• New Framework classifiers should be created if they are "notable;" and
• They will be of immediate use to existing projects (in other words, no creating classifiers simply in anticipation of future uses)

We understand that "notability" is subjective and we'd appreciate help judging that; the information you've already provided goes a long way towards satisfying that.

As to the other guideline ('immediate use to existing projects'), it's been suggested that

• Requestors can demonstrate this by providing links to 10 or more existing PyPI projects that have expressed a desire to use the classifier; or
• Another way to demonstrate this would be to have 10 or more other project maintainers submit comments on the requesting issue about their desire to use the classifier.

We're looking for some feedback on this. Does it seem reasonable to you? And if so, are gathering those links and/or comments something you would be able to help with?