course
Time Series Analysis

Apr 2018

Overall, my suggestion is to avoid this class (Time Series Analysis) if your intention is to learn useful knowledge and not to waste your tuition. Prof. Safikhani is a generous curver but not so generous as a teacher/professor. He constantly cancelled his office hours due to "conferences", which happened at 7:40 at night? His substitute had no idea what he was talking about and he was not even able to read the PowerPoint proper. Prof. Safikhani asserted that he would not answer any homework questions and would left those to the TAs. Unfortunately, we had appalling lady this semester as our TA, who possessed a hateful attitude towards students in this class. So you really should have a solution to do the HW problems, some of which were never covered in class. It is not surprising that the STAT department is packed with irresponsible faculties. Prof. Safikhani is not the worst, but nor could he be called a good professor. His lecture is wishy-washy and offers no insights to the textbook - in fact, it is just a simplified version of the textbook. He is not able to explain essential proofs. The class is very dull. That said, he is a generous curver - he said he would curve the average to an A-. So if your ultimate goal is for a facade of good grades, he should definitely be in your consideration. Yet, I would say such a grade means nothing more than a letter. At the end of the day, you, as a student, are left confused, helpless and disappointed, having paid so much for a useless class with unhelpful TAs and faculty. So save your money and stay away from this.

Nov 2016

This class requires a strong background in probability, calculus and linear algebra. The materials themselves aren't too hard and the lectures are mostly based on the textbook. The lectures themselves can be dry though and the lecture time for this class has always been in the evenings. Stay diligent with the homework as it is long and comprehensive and you will do well in this class. While both undergrads and grads attend the same class, there will be two different curves for each - which will help. He might switch to allowing some of the data modeling to be done in R but for now, there's a program attached with the textbook to do some of the time series labs in the textbook. The program is painfully limited but it's better than doing proofs for homework.

May 2005

Oh lord, how difficult it is to give this guy a good review. Although Hayashi is supposedly known for his work in financial time series, he comes across as a scatterbrain in class. The course is horribly organized, there is little continuity between topics, and important concepts are often omited for students to read about in one of two textbooks, each with its own notation -- very confusing. His lack of teaching ability is evidenced by the fact that he cannot manage his time well in class. In the first part of the course, he pounds the same concept into submission, but then realizing that there is only half a semester left to go, he jumps to ludicrous-speed leaving the students all but confused. Hopefully Hayashi isn't the only one teaching this class, but I fear it might be this way. The only thing you can do is to hang on and frequently supplement class with reading the textbooks. But the subject material is definitely enlightening even if Hayashi's class isn't.