Fisher information. Cramer-Raos inequality. The most important asymptotic results in estimation theory. Sufficiency. Bayesian statistics. Linear models treated generally with matrix formulation. Specialisation to regression analysis, variance analysis and covariance analysis. Estimation and hypothetis testing in linear models. Estimability. Optimality in connection with linear models. Multi-variable distributions from a matrix formulation. Theory on model reduction, prediction and classification.

The students will aquire a basic theoretical understanding of the most important classes of mathematical-statistical models used for instance when analysing biological data, and for the statistical methods developed in connection with such models. They should be able to see connections for exponential distribution classes in general and for linear models in particular. They should also understand why and when a model reduction produces better result. The students will also be able to present subject-relevant material both orally and written.

Lectures, self-study, calculation exercises, project work. The project should be written in a correct scientific language.

The teacher will be available  also on e-mail (thore.egeland@nmbu.no).

Will be given on the first lecture.

STAT100 or equivalent, MATH130 or equivalent.

STAT200 or STAT300, STAT210 or STAT310, STAT321 (or similar courses).

Project assignment.

Written exam, 3,5 hours, counts 100 %.

Special requirements in Science

4 hours per week. Mainly lectures, but also some calculation exercises.

An external examiner controls and approves the examination assignment and grades the examination papers together with the lecturer.