098425 - SEMINAR IN TIME SERIES

Fall semester, 1999

• Instructor: Robert J. Adler
  
  
• Time: Wednesday, 09:30-11:30 (This can be changed if there is demand and agreement. In any case, the first meeting will be on Wednesday, October 20, 1999.)
  
  
• Place: Bloomfield Building, Room 319.
  
  
• Content: The course will concentrate on "model building", and will centre around a number of models not treated in basic courses in Time Series. The issues to be treated include non-linearity and heavy tailed series. These models arise in a number of scenarios, including financial modelling, and the study of the structure of information flow across the Web. I do not plan to go into details of the specific applications, but rather give the tools needed to handle them. However I imagine that some students may wish to look at specific applications as part of their final presentations.
  
  
• Structure: Although the formal structure of this course (in the Technion Catalogue) is that of a "seminar", in which the lectures are presented by students, it will really be an advanced course: i.e. All but the last two or three weeks of lectures will be given by the instructor.
  
  
• Grade: The final grade will depend on presentations given by students at the end of the semester, as well as class participation throughout the semester. There will also be a few homework exercises to give you something to do in the evenings.
  
  
• Prerequisite: The formal prerequisite for this course is 096425 Time series and forecasting. I shall not require all the material from this course, and will spend the first week or two reviewing what is necessary. However, students who have not taken 096425 will need to have taken a course in Stochastic Processes at the level of 098413
  
  

---

COURSE OUTLINE

The following outline is not cast in stone: I will probably change it depending on who takes the course and what we turn out to be interested in. But it should give you an idea of where we are heading. The "core references" will be heavily supplemented by journal papers, and are meant to give only a rough idea of each topic.

Weeks	Topic	Core reference
1	Review of ARMA time series modelling	Brockwell and Davis Time Series : Theory and Methods, Springer, 1991.
2	Review of spectral analysis	Brockwell and Davis
3-5	Stable and other infinite variance models	1: Samorodnitsky and Taqqu Stable non-Gaussian random processes, Chapman & Hall, 1994. 2: Adler, Feldman, and Taqqu, A Users Guide to Heavy Tails, Birkhaüser, 1998.
5-6	Series with long memory, and the connection to heavy tails	Beran Statistics for Long-Memory Processes, Chapman & Hall, 1994.
7-8	Non-linearity via ARCH, GARCH etc.	Engle ARCH: Selected Readings, OUP, 1995.
9-11	Non-linearity via neural nets	Weigend and Gershenfeld Time Series Prediction, Addison-Wesley, 1993.
11-12	Time "series" in the plane	Guyon Random Fields on a Network, Springer, 1995.
13-14	Presentations by students	Journal papers

---

ADDITIONAL INFORMATION

If you want extra information, you can reach me in the office at 8294503, at home at 8251794 (but not Shabbatot or Hagei Yisrael), or, most reliably, at robert@ieadler.technion.ac.il. (Please do not send HTML generated e-mails.)