MATH3821: Statistical Modelling and Computing

- 2 x 10% assignments
- 20% mid-term (laboratory test)
- 60% final exam

MATH2831/MATH2931. It then gets crammed very quickly in the very first lecture as a refresher.

This course is compulsory for all students pursuing a statistics major. (There is no higher counterpart.) It's essentially the course aimed at developing and training real-world application of statistics. The R software is used.

I think everyone felt this way. The lectures were far too packed. A bit too much content was present in this course and getting lost following the lecture was an unusually more common thing here. This change was implemented last year, but apparently there was more advice on how to deal with it. It's a huge trek trying to study for it otherwise.

Perhaps, a bit too much breadth given the time? I think he did try to put depth in, but it was a bit foreshadowed. Regardless, I wouldn't imagine a thing called keeping up with this course is possible unless somehow you've seen it all before.

Biggest drain was in the second assignment honestly. I appreciated the lecturer's genuine marking a lot, but mixing in both a report and the presentation was a bit torturous. Still though, there definitely were improvements - I was very thankful to find that Q&A panel and peer review was taken out. (I really shouldn't say this as a statistics student, but I wasn't interested in hearing about various investigations - I just wanted to do the task!)

Also unlike the first assignment, groups are randomly selected by the lecturer for the second. Some of my mates got put with bad teammates, which made things hard. I got lucky here.

Strong emphasis as in a prior review though - this course is NOT 100% computing. Although both assignments and the midterm revolved heavily around computing, the finals is still a 60% weighted exam, of which only roughly 10% was computing based. It's a math course, so expect some level of theory involved. Computing is just a means to an end when it comes to modelling.

Final exam questions are niche. Doing the tutorials helps a lot, because about half of our questions were based around them. But also you need to understand the content physically to be able to do a large portion of the rest of the exam.

It's just that proofs weren't really assessed much until the finals. The only proof question in assignment 1 was straightforward.

Also: Although MATH3871 is definitely not assumed knowledge, I found that doing it beforehand made the Bayesian half of this course FAR more dealable. Whilst some of my peers were forced to learn Bayesian altogether, I was like "oh yeah this is just 3871 gone nuts". Could be advice for you? Keep in mind though Bayesian is now offered in T3, whilst this is a T2 course.

3 hours lecture in total, 1 hour lab, 1 hour tutorial

4/5

Yes

Tom Stindl

Slides are comprehensive but arguably dense...
Tutorial and laboratory questions uploaded, but solutions were only uploaded to the ones not covered in class. No real other resource.

3/5 3.5/5 - Experience made better following paper inspection

None, but this one was recommended as a side resource.

19t2

94 96 HD

2 * 10% Assignments, 20% Mid-semester (Lab) Test and 60% Final Exam

MATH2831/MATH2931 is prerequisite

If you think this is an easy computing course, then you've walked yourself to the wrong door. MATH3821 is a *

2 Hours of Lecture, 2 Hours of Laboratory

Depends entirely on your effort, but I would say it is pretty easy 2/5

Yes

Dr. Pierre Lafaye de Micheaux

N/A

4/5

N/A

2018/2

HD