Note: Copied from my response in the Gitter chat room
I like the dichotomy between data science and “machine learning”/“statistical learning”. Although I believe that machine learning is a subset of what goes on within data science (which I believe is @evaristoc’s argument for keeping them under one tag), others outside have quite a distinction for those focusing on just machine learning algorithms (e.g. testing out algorithms on Kaggle) and all the rest of data science people don’t talk much about (e.g. data cleaning).
Although this separation is not fair on the other important topics of data science, I think people have interests in machine learning specifically that it may warrant its own tag. And I bet this is probably what @QuincyLarson was thinking when he listed them separately in his forum post proposal.
At the end of the day, I feel the bigger question may be what is the pros and cons for having one versus the other. One benefit for separating the two is to have a more focused tagging of forum posts, which I think would be machine learning and everything data science non-machine learning. However, now you have two tags, which are really just subsets of each other. So if you have one tag, it’ll be easier to classify everything as just one tag of “data science”. But then you have a problem of searching well if you go by tags. Eh, but I guess the search functionality of Discourse should be good enough to catch various search terms?
tl;dr I like splitting data science and machine learning into two because external forums/the internet I feel tends to lump into these two groups in terms of frequency of discussion