Göteborgs universitet
Matematisk statistik
PARAMETERFRIA METODER (Anm kod: MSN10)
A 5 point undergraduate statistical course based on the textbook
"Nonparametric Statistical Inference"
by J.D. Gibbons and S. Chakraborty, 3rd ed., Marcel Dekker, New York,
1992 (can be purchased at the Cremona bookshop)
starts on 20 January 2000 and takes place on
Tuesdays and Thursdays at 9.00-11.45, room MD7, MC.
Prerequisite: basic courses in calculus, probability and statistics.
The major topics of the course:
- order statistics
- tests of randomness based on runs
- goodness-of-fit tests (chi-square, Kolmogotov-Smirnov, Lilliefors)
- rank-order statistics, treatment of ties
- sign tests, confidence intervals for median (Wilcoxon, Walsh averages)
- two-sample tests for the location problem (control median, Mann-Whitney)
- two-sample tests for the scale problem (Mood, Siegel-Tukey)
- one-way layout tests (Kruskal-Wallis, Jonckheere-Terpsta)
- non-parametric mesures of dependence (Kendall's tau, Spearman's
rank correlation)
- two-way layout tests (Friedman, Page, Durbin)
- asymptotic relative efficiency of two tests
The term "nonparametric"
assumes a general statistical setting when the underlying population distribution does not
belong to a certain family. We will study these techniques keeping in mind the parametric
methods that inspired them.
Formula list in the Acrobat-format.
Instructor: Serik Sagitov (room
1420; tel. 772 53 51; serik@math.chalmers.se; http://www.math.chalmers.se/~serik/)
Grading system
Students regularly will be asked to solve certain problems from the
book.
To pass a student should collect min. 14 units while the ``VG'' threshold
is 23 units out of 33 ( = 30 for a written exam + 3 units bonus for home assignments).
Summer 2000 exam holds on Tuesday Aug. 22, 8.45-13.45 in the MG building.
Last modified: Thu Aug 24 16:43:50 MET DST 2000