Gabriel Young

Homework: homeworks are LONG. There are a lot of questions. Distributed on bi-weekly basis. But I think they are important for learning the tricks, although they are typically not conceptually interesting. It is advised that you do all these homeworks because the exams are under non-trivial time pressure (~17 problems, counting all the sub-questions that should really be an independent question, under 90min). You want to be very familiar to be able to ace each problem <5min and to actually finish the exam in time (I think he designed the exams so that students are NOT expected to finish). Why is 5min a significant time pressure? For some reasons, he really likes throwing out computationally cumbersome problems... Grading: Generous curves--if you are above the median, you get an A-. <= 1~2 A+s. I think he's been toning down on the exams over the years. There were some very hard questions in earlier years' exams, but this year most questions are reasonable. However, the problems proliferate in quantity. So again I suspect that the challenge going forward is to finish the exam... Overall I think there's no need to worry about grades. The class tends to have bi-modal distribution so that the exam median is always somewhere around 55%. Which means if you can keep up with the class you will do fine and get an A.

The theoretical foundations of probability starting from set theory and counting techniques to multivariate extensions of probability density functions.

AR, MA, ARMA, Y-W Equation, Innovation algorithm

The theoretical mechanics of statistical machine learning.

Very good course the combines theory and data. Wish there was slightly more focus on the math (maybe optional hw tracks or a different project for students who prefer math over data?) e.g. we did not talk about spectral methods of time series analysis

A laundry list of important machine learning topics, really difficult to name all of them because every class covers something new.

Intro to R with respect to the field of data science.

Midterm, Final, bi-weekly homework. Not bad-- the homework can get a little time intensive but if you distribute it over the two weeks that you are given to do it then it isn't bad at all.

Moderate, if you have coded before this class really isn't too bad. - There's 1 in class lab a week (attendance was mandatory for these days) and 1 homework every 2 weeks. (Pretty much everyone gets 100s or high 90s on these, the last couple assignments are a little tough but you can go to office hours for help) - 1 midterm - 1 final

Extremely solid instructor, very effective teaching. I think this is as good as it gets for a fast-paced, non measure theory based introduction to probability (and a little bit of statistical inference). Covered all the essentials of probability theory, up to convergence results (Central Limit Theorem, Weak Law of Large Numbers, Convergence in probability vs. in distribution, etc.), the same as one would get out of Stat 4203 (minus some distributions, like beta, but computationally this is just about knowing a few more formulas, nothing fancy). But warning is that the class is quite dense, and the pace becomes especially fast entering the second half (joint distributions, etc.). So it's definitely non-trivial. And a good review of various integration techniques, single and multivariable (or pick them up if you haven't already learnt them). I feel like 80% of this class is familiarizing oneself with integration tricks that cater to probability theory scenarios. And only 20% is about learning concepts more specific to developing probability theory. But I guess this is exactly what one should expect from an applied probability class. (And who's to say that a rigorous course in probability is not 80% about applying analysis and measure theory to probability settings...Alas) And he's extremely nice and genuine and fun to talk to. Very chill person. Def take a class with him if you…