Matematik / Matematisk statistik 


Course Log SF2943 Time Series Analysis, Spring 2014
  • 2014-09-02. That's all for this year.
  • 2014-09-02. Grades were set accordning to the total score (exam and bonus):

    F: 0-21, Fx: 22-23, E: 24-28, D: 29-33, C: 34-38, B: 39-44, A: 45-60.

    Exam statistics: F: 9, Fx: 0, E: 3, D: 2, C: 10, B: 0, A: 5.

  • 2014-09-02. I have evaluated the exams. Solutions to the exam problems can be found by following the "old exams" link on the course webpage.
  • 2014-06-11. Grades were set accordning to the total score (exam and bonus):

    F: 0-22, Fx: 23-24, E: 25-29, D: 30-34, C: 35-39, B: 40-44, A: 45-60.

    Exam statistics: F: 15, Fx: 3, E: 7, D: 10, C: 5, B: 10, A: 6.

    For those of you that got Fx: contact me immediately to enable a grade to be reported before the end of the summer.

  • 2014-06-10. I have evaluated the exams. Solutions to the exam problems can be found by following the "old exams" link on the course webpage.
  • 2014-06-10. I would very much appreciate if you give me your opinions on the course by spending 5 minutes on the course evaluation.
  • 2014-05-26. Office hours: Wednesday 28 May 10-12 (Pierre), Monday 2 June 10-12 (Filip). Send an email to make an appointment.
  • 2014-05-22. Bring a pocket calculator and Formulas and survey, Time series analysis by Jan Grandell (without notes!) to the written exam. Information about the location of the exam is found here.
  • 2014-05-15. Presentation of selected problems from old exams. There are no more lectures or assignments. However, I will use this course log for information to you, so keep following updates during the weeks ahead.
  • 2014-05-15. Some notes on ARMA processes are found here.
  • 2014-05-14. Note that we meet in M33 on Thursday morning, i.e. not where we usually are.
  • 2014-05-14. I gave an overview of state-space models and Kalman recursions. More information on these issues can be found in the lecture notes by Timo Koski, avaliable here.
  • 2014-05-13. I presented ARCH and GARCH processes and explained the connection to AR and ARMA processes.
  • 2014-05-12. I have evaluated the reports. I used several criteria including correctness, quality of presentation, presence of insighful comments and conclusions, demonstration of ability to select the most appropriate and relevant methods. I will bring the reports to the lecture on Wednesday.
  • 2014-05-12. A short remark on the presentation of problems from old exams on Thursday: the group submits a short report (at most one page) printed on paper during the lecture on Wednesday. If you cannot (or do not want to) attend the lecture on Wednesday, then you may send it by email to Pierre and me before the lecture on Wednesday (subject line: "old exams"). Based on the collection of short reports I will ask a few groups to present solutions to the selected problems on Thursday.
  • 2014-05-08. There were several very nice reports handed in. One of them is the following report by Benjamin Kjellson. Thanks Benjamin!
  • 2014-05-08. Pierre talked about Problems 2, 3 and 5 from the May 2012 exam, Problem 2 from the May 2013 exam, and Problem 5 from the June 2011 exam.
  • 2014-05-07. I talked about h-step prediction for AR(p) processes, forecasting with multivariate AR(p) processes (a bivariate example with financial return series), and cointegration. I made some introductory comments on financial times series.
  • 2014-05-06. Project presentation seminar. Nice presentations of very good project work. Information on the data sets analyzed in Problem 6 can be found here.
  • 2014-05-05. Pierre and I have read your reports. Most reports are nice to read and present good solutions to the problems. The reports also give relevant feedback on issues that we should clarify. We need some more time to take a closer look at those issues. Therefore, you will not get the reports tomorrow, but soon.
  • 2014-05-05. I have asked a few groups to present their work during the project presentation seminar tomorrow. No preparation is required for the presentation seminar besides being prepared to explain and present your work to the class.
  • 2014-04-28. Pierre presented Example 5.4.1, Problems 5.1 and 5.4, and Problem 1 from the December 2010 exam.
  • 2014-04-24. The second project discussion seminar took place. Pierre and I heard many interesting discussions, good ideas and creative solutions. Great!
  • 2014-04-23. You have the opportunity to visit Pierre on Friday morning (8-11:30 AM) to discuss time series issues that you struggle with. To visit Pierre you should send Pierre an email to make an appointment.
  • 2014-04-23. I talked about model diagnostics (how to create something that should be a white noise sequence if the fitted model is the right time series model for the data) and order selection for ARMA processes.
  • 2014-04-22. I talked about parameter estimation, Yule-Walker, the Innovations Algorithm, and Maximum Likelihood estimation.
  • 2014-04-17. Pierre presented Exercises 4.4, 4.5, 4.6.
  • 2014-04-16. I talked about spectral densities and periodograms, and used R to plot spectral densities and draw conclusions from the graphs. A short summary, including some R-code, is found here.
  • 2014-04-15. I presented the results of the multiple choice test and discussed how to solve the problems.
  • 2014-04-10. Comments on the multiple choice test results:
    96% passed the test. 85% submitted answers prior to 9th April and 100% of those passed the test (a few needed to send a new set of answers). The median of the test scores was 8. The incorrect answers provide me with relevant information on what topics should be discussed in more depth. From the project discussions this morning I got the impression that you are well prepared to do good project work. I believe and hope that the multiple choice test has contributed to making you well prepared for the project work.
  • 2014-04-10. Correct answers to the multiple choice test problems:
    D A B A D C B C D D
    We will discuss the test problems and the results later.
  • 2014-04-08. I will not communicate the correct sequence of answers to the multiple choice test, I will only let you know whether you passed the test or not. Later in the course, we will discuss the multiple choice test.
  • 2014-04-08. Today I talked about forecasting (the best linear predictor), the partial autocorrelation function, its usefulness for order selection with AR(p) models, and estimation of the mean of a stationary time series.
  • 2014-04-04. A typo was found in Problem 10 of the MC test: alpha(0),alpha(1), ...,alpha(4) should be alpha(1), ...,alpha(4). A new version is available.
  • 2014-04-04. Yesterday Pierre presented Exercises 2.9, 3.1, and 3.4. The extensive lecture notes of Jan Grandell are available as a pdf file from the course webpage. There will soon be a printed version available too (at the students' office of the department of mathematics).
  • 2014-04-01. The multiple choice test is now available. Note that I changed the deadline on 9 April from 13:00 to 15:00.
  • 2014-04-01. For students from SU without a KTH-id: registration for the exam can/should be done at the students' office of the department of mathematics.
  • 2014-03-27. Project instructions are now available.
  • 2014-03-27. Pierre presented Exercises 1.1, 1.4, and 1.8.
  • 2014-03-25. I presented some basic analysis of sea level measurements, summarized here and based on this data set, that gave an introduction to some topics in the course. Try it out yourselves!

Last change 2014-06-11
[Kurshemsidan]     [Kursförteckning]     [Avdelningen Matematisk statistik]
Sidansvarig: Filip Lindskog
Uppdaterad: 2014-03-17