The
Tenth Course in the ECAS Programme:
Regression quantiles
and applications |
La Roche-en-Ardenne,
Belgium |
September
12 - 16, 2005 |
Scientific
Programme Committee
J. Beirlant, Katholieke Universiteit
Leuven, Belgium
J.-J. Droesbeke, Université
Libre de Bruxelles, Belgium
M. Hallin, Université Libre
de Bruxelles, Belgium
S.Heiler, Universitat Konstanz, Germany
H. Oja, University of Tampere, Finlande
L. Simar, Université
Catholique de Louvain, Belgium
Organising
Committee
D. Cassart, Université Libre
de Bruxelles, Belgium
C. Dehon, Université Libre
de Bruxelles, Belgium
J.-J. Droesbeke, Université
Libre de Bruxelles, Belgium
M. Hallin, Université Libre
de Bruxelles, Belgium
D. Paindaveine, Université
Libre de Bruxelles, Belgium
C. Vermandele, Université
Libre de Bruxelles, Belgium
Scope
of the course
Quantile regression is a
statistical
technique intended to estimate, and to conduct inference about,
conditional quantile functions. Just as classical linear regression
methods based on minimizing sums of squared residuals enables one to
estimate models for conditional mean functions, regression quantile
methods offer a mechanism for estimating models for the conditional
median function, and the full range of conditional quantile functions.
Estimation here is based on a weighted sum (with weights depending on
the order of the quantiles) of absolute values of residuals. By
supplementing, the conditional mean with an entire collection of
conditional quantiles, regression quantile methods provide a much more
complete statistical analysis of the stochastic relationships among
variables; in addition, they are more robust against possible outliers,
and can be computed via traditional linear programming methods.
Although median regression ideas go back to the 18th century and the
work of Laplace, regression quantile methods were first introduced by
Koenker and Bassett (1978) in a seminal Econometrica paper. Since then,
they have generated a huge literature, and have found innumerable
applications.
Along with the dual methods of regression rank scores, regression
quantiles can be considered one of the major statistical breakthroughs
of the past thirty years.
Topics
- Introduction.
R. Koenker (University of Illinois, USA)
- Fundamentals.
R. Koenker (University of Illinois, USA)
- Asymptotics.
I. Mizera (University of Alberta, CANADA)
- Rank Based Inference.
B. Werker (University of Tilburg, NETHERLAND)
- Introduction of QR Inference.
R. Koenker (University of Illinois, USA)
- Nonparametrics.
S. Heiler (University of Konstanz, GERMANY)
- Penalty Methods for Nonparametrics.
I. Mizera (University of Alberta, CANADA)
- Longitudinal Data Analysis.
R. Koenker (University of Illinois, USA)
- Multivariate Analysis and Data Depth.
I. Mizera (University of Alberta, CANADA)
- Time-Series Analysis.
B. Werker (University of Tilburg, NETHERLAND)
- Survival Analysis and Binary Response.
R. Koenker (University of Illinois, USA)
- Tutorial on Computational Methods.
R. Koenker (University of Illinois, USA) and I. Mizera (University of
Alberta, CANADA)