Building ML models/visualization for categorizing users into "buckets"

[iantei]

"On the "building models/visualization" side: especially as we have these larger, 100+ user deployments, it would be good to group the users into "buckets" that the admins can quickly check. The new metrics that we just added (and other new metrics that come up) can serve as features into a basic clustering model, or we can try to come up with categories ("mostly passive", "active labeler") and treat it as a classification problem. We might also want to incorporate checks that more sophisticated admins (like Matt from UNC) have already implemented internally."

• @shankari