Messages
Slides for the summary is now available under the "lecture plan"
A summary of the course will be given tuesday november 25 at 12.15-14.00. I will put a pdf file containing part of what I will talk about on the webpage tuesday morning.
Exam The exam is a written project followed by a written exam on Friday December 5th 2008. The project paper and the exam counts equally and the final grade is based on ? general impression after the final exam. (The two parts of the exam will not be indivdually graded.)
Permitted aids to exam: Approved calculator.
The final syllabus list is now available at the "syllabus" link
Project: Should be handed in IN TWO COPIES to the expedition of dept. of mathematics, 7th floor, math building.
Project: Some of the methods we have discussed have problems with categorical covariates. However, since all the categorical variables in the project data only have two levels, you can equally well treat them as numeric.
The project, which is part of the course assessment is now available from the "project" link at the left of the homepage.
Because of lack of time, chapter 10 will not be covered in the course.
Extra exercise 12: You will need to give the command library(mgcv) before calling the gam function.
We include the first part of ch 5 since some of this is needed in ch 9.
As pointed out by Audun M. ?ygard , Extra 8 is (almost) identical to Exercise 3.6 from the textbook. Concentrate therefore on extra 9. I will see if I am able to find an additional exercise as well.
Changes in lecture plan: We skip ch 8, and rather jump to ch 9 when we have finished ch 7.
Audun M. ?ygard has corrected me on the use of the deg option in the loess routine in R in that in order to get the Nadaraya-Watson estimator, you use deg=0 while local linear regression corresponds to deg=1. I have now changed all the R-scripts on the web in accordance to this. Appologize for my misunderstanding of this.
Room B70 in the 7th floor in Niels H. Abels house (math building) is now reserved tuesdays 9.15-10.00 for STK4030. Students can come there and work with the weekly exercises, preferable with a lot of interaction! There is also a terminal room in the 8th floor. I will be in my office (B836 in the 8th floor) at this time and will be available for questions.
Exerise 6.12 turned out to be a bit more difficult than expected. A code is therefore now available from the "Exercise" link
There was a typo on extra 4 a and in 4 c. These have now been corrected.
The course assessment will be through a practical project and a written exam. The project will be available from Nov 4 and has to be handed in at Nov 21, 12.00. The written exam will be Dec 5.
Examination/form of assessment: A project followed by a written exam.
The code needed for producing table 3.3 turned out the be a bit more complex than I expected. I will therefore in a short while provide you with some code you can use.
On Monday August 25, Robert Tibshirani (one of the authors of our textbook) will be in Oslo and give a talk. This is an excellent chance to listen to one of the most influential researcher within this field. He is the main inventor of the Lasso method, which we will discuss in the lecture in a few weeks (page 64 in the textbook) and his talk will probably include some about this. Even though you perhaps will have some difficulties in understanding everything in the talk, it will give an impression on what is going on in research in this field. The talk will be Monday August 25, kl. 14.00-15.00, Domus Medica, room Nye Auditorium 13 (see map, nr. 19, first floor)
If you have problems getting the book from the bookstore, note that it possible to order the book on the net, see thebook webpage
Slides from first lecture is now available from the "Lecture plan" page. For the next lectures, I will use the blackboard.
On Tuesday August 19 there will be no excersises so the teaching will start at 12.00.