TAMS69	Computerintensive Statistical Methods, 3 ECTS-points /Datorintensiva statistiska metoder/ Advancement level: D
	Aim: Computer-based methods give revolutionary new possibilities to evaluate predictability and estimation uncertainty in model selection, regression, and time series situations. This is today an active research area. Principles and applications are illustrated and the students are taught to understand and apply some of the new tools. Prerequisites: Basic course in probability and statistics. Multiple regression. Familiarity with computers and some language like Fortran or Pascal. Time series and simulation are helpful but not necessary. Course organization: The course is given as 12 times 2 hours seminars in class and 6 times 2 hours computer exercises at work stations where usually two persons work together. The seminars and exercises interfere quite a lot in order to cover a rather large material in relatively few hours. Course content: Models. Computer intensive philosophy. Strange statistical effects of model selection. Cross validation and cross model validation. Validation of model selection. Forward validation of time series. Bootstrap methods. Bootstrap analysis of parameter estimates. Bootstrap confidence intervals. Applications on data from e.g. traffic, meteorology, stock market, reliability, medicine etc. The applications can vary somewhat between different years. Course literature: Hjorth, U.: Computer Intensive Statistical Methods, Chapman and Hall, London 1994.
TEN1	Written examination, 2 p.
LAB1	Laboratory work, 0,5 p.

Course language is swedish.