BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Dimension reduction based on sliced inverse regression (SIR): a lo
 ok at the special case when n ‹p  - Jérôme Saracco\,  Bordeaux Institu
 te of Technology &amp\; Bordeaux Institute of Mathematics (Probability and
  Statistics team)
DTSTART:20121120T143000Z
DTEND:20121120T153000Z
UID:TALK41492@talks.cam.ac.uk
CONTACT:Dr Jack Bowden
DESCRIPTION:In this talk\, we first give an overview of sliced inverse reg
 ression (SIR) which is an attractive dimension-reduction approach to model
  the effect of the p-dimensional covariates x on y via a semi-parametric r
 egression model. Several authors proposed and studied SIR-based methods wh
 en the sample size n is greater than p. These approaches do not work when 
 n‹p since they are based on the inversion of the variance matrix of the 
 covariate x. Then\, we present some procedures to tackle this issue and we
  compare them on simulated data. 
LOCATION:Large  Seminar Room\, 1st Floor\, Institute of Public Health\, Un
 iversity Forvie Site\, Robinson Way\, Cambridge
END:VEVENT
END:VCALENDAR