Fall 2013 2.c
The model solution feels like it no longer relates to Goldburd, which is focused on understanding and interpreting GLMs instead of solving them. I would not have thought to take partial derivatives to solve (at least, not without brushing up on MAS content 😊).
Is this still a valid question on the current exam?
Comments
This is very hard to say given the CAS isn't releasing exams and candidate feedback to us and in general is restricted. The CAS does say they can draw on any material from prior exams - though this seems like a really niche piece of knowledge given even if you're working with GLMs day-to-day the computer is doing this bit for you.
I would say it is less likely but not off limits. Focus on the broad theme: how to set up the design matrix, write out the equations with the error terms and understanding we're trying to minimize the sum of squared errors. From there I'd worry about it on exam day if it comes up and focus now on other material that's more likely.