Statistics 516/517 : Stochastic Modelling of Scientific Data
Considering these are now the core required statistical courses for our program, it's probably time to add some info.
[Note that I just started 517 so I'll have to come back and fill this out later.]
Naturally the first place to start is the official course description as well as Professor Minin's description:
- "The purpose of this course is to introduce students to the art of stochastic modeling. The theoretical component of the course covers material standard for a first course in stochastic processes. However, emphasis on statistical inference and scientifically motivated examples give a unique flavor to the mathematics presented in the course. The first quarter of the Stochastic Modeling sequence will be devoted to discrete and continuous-time Markov chains on countable state spaces."
Professor Vladimir Minin, who typically teaches the courses, requests that we do not post his lecture notes or homework solutions on the Wiki. However, I got permission to add the homework problems and practice exam questions. The homeworks (for Fall 2010) are merged into one file here and the practice problems (not necessarily actual exam questions, just good practice) can be found here. These two documents should give a good idea for what material is actually covered.
STAT 516 typically had a homework assignment due each week, and a midterm and final examination. STAT 517 has a slightly different format, in that there are fewer assignments and a large "project" at the end of the quarter. From what I understand the project is a literature review on a paper of choice.
In our cohort there was not a lot of mathematical statistics background, and I would recommend working through some of those concepts before starting the class. For me in particular maximum likelihood estimation and Bayesian inference would have been very helpful, so that the "review" is actually that. In addition a better understanding of some of the limit theorems (SLLN, CLT) would have been nice preparation for the build-up to the Ergodic theorem. I never took STAT 481, but it looks like that would be a good introduction. I did take STAT 394/395 and found those to be very helpful.
In terms of programming we started off with some basic simulations and built up to pretty involved MCMC by the end of the quarter. So some previous experience coding will be helpful. Note that he allows you to use any high-level language, but as a qermie R should be an obvious first choice.