Messages

As previously explained & agreed upon, " You are allowed to bring *one sheet of paper with your own handwritten notes* (in as tiny a handwriting font as you might wish), but otherwise no book and no other material. " translates to handwriting on one side of the physical cellulose-derived product in question, not both sides.

In addition to your Exam Project we have the four-hour written examination on Monday December 8, 9:00 - 13:00, in Store fysiske lesesal Fysikkbygningen. You are allowed to bring *one sheet of paper with your own handwritten notes* (in as tiny a handwriting font as you might wish), but otherwise no book and no other material. The exam is not a "memory test", and if a point or two might need a special formula, I'll simply give it.

Bring also a simple pocket calculator (but nothing with a big memory capacity, since the "exam guards" then will grow professionally suspicious) -- you might need som simple calculations, involving a square root or an exp or a log or the like, but essentially nothing more.

Good luck & happy work.

Nils Lid Hjort

The exam project is ready for you, check under "Exam work December 2014" on the right hand side of this page. Remember to fill in the declaration form "page A" as well as to include a "page B" self-assessment statement when you hand in your reports (in duplicate, please). Good luck & happy work.

For this Wednesday we go through some of the exam questions from 2013 (both "the project" and the four-hour written exam), and also use the opportunity to look back at the more crucial issues from the curriculum. This is also a chance for "questions and answers". We use an extra hour, in B81 from 13:15 to 14, for those interested. The exam project will be made available on the course site on *Tuesday December 2*. Reports are to be handed in by *Thursday December 18*; further practical details regarding this will be made available separately.

From Ch 8: 13, 15, 19.

From Ch 9: 2, 3, 4, 37.

1. On Wed 29 October I discussed the last parts of Ch 7 and went through various exercises. Next week we start Ch 8. Not all parts of Chs 8-9-10 are inside the curriculum, cf. the list given here some time ago, and we shall find time at the end of the course for "repetition" and exam preparation work.

2. Exercises for next week: Ch 8: 1, 2, 3, 6. Here you will need to check with the start of Ch 8 yourselves, regarding e.g. the likelihood ratio test, even though I have not yet really included this in my lectures -- so take these Ch 8 exericses as a kickstart. Furthermore: do Ch 6: 37 and Ch 7: 34, plus Ch 7: 37, 46, 47.

1. On Wed 22 Oct we discussed Cramer-Rao and ML, along with more on sufficiency and method of moments.

2. Exercises for Wed 29 Oct, Ch 7: 11, 14, 19, 20, 21, 24. But first do this extra exercise: (a) For iid observations from the uniform [0, theta], let M_n be the largest observation, and consider \hat\theta = a M_n. Which value of a is best, judged from mean squared error mse(\theta)? (b) For iid observations from the normal (mu, sigma^2), consider \hat\sigma = b S, where S is the usual empirical standard deviation. Which value of b is best, based on mean squared error?

3. Thu Wed 30 I give a talk at the "Faglig-pedagogisk dag", "Skisprint: Semifinalenes betydning for medaljene". It might contain exam relevant modelling & inference.

http://www.uio.no/om/samarbeid/skole/fagped-dag/program/?fagkode=MAT

1. Today Wed 15 October I went through the first half of Ch 7, including method of moments, method of quantiles, a bit of maximum likelihood. I continues with these themes next week. We did not manage to get through all exercises, but this will be partly accounted for next week, and the planned "extra hour of exercises" two Wed 29 October.

2. Note that "Exercises and Lectures Notes" (version A) are now available at this course site. Exercises for Wed 22 October: First, #3 and #10 from this extra set. Next, do Exercise 7.2, with cuckoo egg sizes data from 7.10, and use a suitable computer programme to fit the gamma(a,b) density (with mean a/b and variance a/b^2), using (a) maximum likelihood, (b) method of moments, and (c) method of quantiles, using the 0.2 and 0.8 quantiles. Plot the three fitted gamma densities in a diagram, and comment. Furthermore, from Ch 7: 6, 7, 9.

I am in Bergen at the Johan Hjort Symposium, and G.H., my intended lecturer & teacher replacement for Wed 8 October, unfortunately cannot make it after all. PhD stipendiats Céline Cunen and Aliaksandr Hubin will therefore experience their Leonard Bernstein moments, as they will take responsibility for going through the exercises (nos. 1, 2, 3, 7, 12, 13, 17 from Ch. 6), with the intention of accomplishing this from 9:15 to 11:00 (i.e. they might not need the full time window 9:15 to 12:00).

The curriculum is based on Casella and Berger's "Statistical Inference" (2nd ed., 2002), and is detailed below. Also, all exercises discussed during the course are to be seen as part of the active curriculum, including those from the "Nils Exercises and Lecture Notes" (made available here at early October). You should look through material given as "cursory curriculum", attempting to grasp the essence of it, but it is not "active curriculum" in the direct exam preparation sense of the word.

Chs 1-2-3: Cursory.

Ch 4: You should know and be able to use the transformation formula, giving the density of Y = h(X) based on the density of X, where X and Y are vectors of the same dimension and h is smooth.

Ch 5: Active, but not 5.6.

Ch 6: Active, but not 6.3, 6.4.

Ch 7: Active, but not 7.2.4.

Ch 8: Active, but not 8.2.2, 8.2.3, 8.3.3.

Ch 9: Active, but not 9.3.

Ch 10: Active, but not 10.3.2, 10.4.2.

Nils Lid Hjort

1. On Wed Oct 1 I went through the last parts of Ch 5, including the delta method, and started on Ch 6.

2. Next week I am at the Johan Hjort Symposium, and Gudmund Hermansen will teach in my place. He will focus on Ch 6 material, specifically Sections 6.1-6.2, and 6.3-6.4 is not part of the curriculum (see separate list). Exercises to be discussed, all from Ch 6, are 1, 2, 3, 7, 12, 13, 17.

http://www.imr.no/om_havforskningsinstituttet/arrangementer/konferanser/johan_hjort_symposium_2014/en

3. In a separate message I list chapters and sections defining the curriculum (pensum) for the course.

1. On Wed Sept 10 I discussed (a) score functions and Fisher's information matrix, in the context of Ch 3, and (b) the multivariate transformation theorem and a few aspects of the multinormal distribution, from Ch 4, before I started with Ch 5. I plan to use the next two weeks for the rest of Ch 5.

2. Within a few days I will write up some brief "Lecture Notes & Exercises" regarding the material pointed to above.

3. The exam for this course consists of two parts -- *a project with a project report* (part A) and *a four-hour written exam* (part B). The date for part B is Mon 8 December. The *tentative dates* for part A are that the project is made available *Tue 2 December* with the reports to be handed in *Tue 16 December*. Please tell me quickly is this for some reason would be practically difficult for any of you.

4. Exercises for Wed 17 September are from Ch 5: 2, 10, 11, 13, 16, 17, 18, 24, 25.

1. On Wed Sept 3 I went through a decent list of exercises, and then discussed various matters from Ch 3 (families of distributions, inequalities, transformations; incidentally, the book's "Chebychev" is a somewhat odd transliteration of Чебышёв). There are just a few more things I wish to point to from Ch 3 before I start with Ch 5, next week.

2. Exercises for Wed Sept 10 are from Ch 3: as many as we find time for, of 4, 6, 8, 13, 14, 20, 25, 26.

3. After being asked about this, I consider it a good idea that we *once in a while*, perhaps once a month, spend one more Wednesday hour, perhaps 12:30 to 13:15, going through more exercises than what we otherwise find time for inside the 9:15 to 12:00 time window, and with more time for details. We'll be coming back to practical details regarding this.

4. We will be determining the precise dates for Exam Part A ("the project") and Exam Part B (the four-hour written exam) shortly.

The course will be taught Wednesdays 9-12 in seminarrom 123, Vilhelm Bjerknes building. The course material will in essence be taken from Casella and Berger's "Statistical Inference" (2nd edition). You may look through messages and information pertaining to the version of this course which was given autumn 2013.

Nils Lid Hjort