This task is somewhat related to "Understanding first day", although it is not explicitly part of that project's scope. We should discuss and think if or when it would be good to work on this.
"Understanding first day" is all about learning what new account holders do between when they create their account and when they make their first edit (or not). One open question is: what kind of edits do newcomers make? Do they make small ones? Do they write new articles? Do they edit popular articles or obscure ones? Are there certain types of edit patterns that most newcomers exhibit?
Perhaps it is possible to look at the content of the edits of a newcomer's first session by using the public data in the user and revision tables, and classify them into segments, like "single small correction" or "new article creation" or "vandalism". It would be additionally illuminating to know which of those first sessions resulted in reverts.
By counting all this up, we would learn more about what new editor activities are common, and what actions lead up to those activities. Ultimately, this would help us guide new editors toward productive contributions.