European Courses in Advanced Statistics - Courses
The Second Course in the ECAS Programme:
| Robustness in Statistics |
| GÜNZBURG, GERMANY |
| October 2-7, 1989 |
Scientific Programme Committee:
S. Heiler (Konstanz, F.R.G., Chairman)
V. Barnett (Sheffield, U.K.)
P. Rousseeuw (Brussels, Belgium)
Ph. Tassi (Paris, France)
N. Victor (Heildelberg, F.R.G.)
Organising Committee:
N. Victor (Heildelberg, F.R.G., Chairman)
J. König (Heildelberg, F.R.G., Secretary)
L. Bretch (Konstanz, F.R.G.)
J. Droesbeke (Brussels, Belgium)
S. Heiler (Konstanz, F.R.G.)
N. Lauro (Naples, Italy)
P.Michels (Konstanz, F.R.G.)
Scope of the course:
Whereas the consideration of outliers has a long historiacal tradition, the idea of robustness of a statistic as the property of performing well also whe the underlying stochastic asuumptions in a parametric model do not hold exactly, was introduced only in the mid sixties by P.J. Huber. Further important contributions led to the concept of influence function by F.. Hampel in 1968 and to exxtensions of the idea on multivariate data, regression and time serie. In the meantime a large stream of contributions has broadened the theoretical basis as well as the tool box for practical applications.
For the concept of outliers, outlier-generating models and test for outliers, the detectionof multiple outliers, performance criteria for test and detection rules were developed.Masking and swamping effects of various methods and the outlier proneness of distributions were considered.
The present this stream of contributions in a structural way is the main purpose of the course. The theoretical foundation will be introduced as well as practical applications with computer software.
Topics:
Basics concepts of robustness theory, M-.estimates and other approaches, including high breakdown point estimates and test for single parameters, parameter vectors and regression; outlier generating distribution models detection and test for outliers; diagnostic from robust regression; issues for time series, robust ARMA fitting, smoothing and spectral density estimation.
Lecturers:
A.C. Atkinson (London, U.K.)
N. Fieller (Sheffield, U.K.)
U. Gather (Dortmund, F.R.G.)
F.R. Hampel (Zürich, CH)
A. Marazzi (Lausanne, CH)
R.D. Martin (Seattle, U.S.A.)
E. Ronchetti (Geneva, CH)
Rousseeuw (Brussels, Belgium)
W.A. Stahel (Zürich, CH)