BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:MRC Biostatistics Unit Seminars
SUMMARY:&quot\;Better prediction by use of co-data: Adapti
 ve group-regularized ridge regression&quot\; - Dr 
 Mark van de Wiel\, VU University Medical Center an
 d VU university\, Amsterdam
DTSTART;TZID=Europe/London:20160322T143000
DTEND;TZID=Europe/London:20160322T153000
UID:TALK65252AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/65252
DESCRIPTION:For high-dimensional settings\, we show how one ca
 n use empirical Bayes (EB) principles to estimate 
 penalties that may differ across groups of variabl
 es. These groups are predefined using co-data\, wh
 ich is auxiliary information available on the vari
 ables (e.g. genomic annotation or external p-value
 s). Due to the adaptive character of the penalties
 \, the group-wise penalties may improve prediction
 s when the groups are indeed informative\, while n
 ot deteriorating those when this is not the case. 
 We provide an implementation in a classical logist
 ic ridge regression setting. However\, we will als
 o discuss extension of the framework to a Bayesian
  ridge regression setting. The latter is particula
 rly useful for obtaining credibility intervals on 
 the predicted event probabilities. In particular a
  hybrid EB-Full Bayes approach in combination with
  highest-probability density intervals seem to hav
 e good coverage properties when the number of vari
 ables is not extremely large. Finally\, the potent
 ial for better variable selection\, either by post
 -hoc selection or by sparse regression\, will be s
 hortly considered. Several real data examples will
  be discussed\, in particular on cancer diagnostic
 s using a variety of molecular data types\, such a
 s methylation\, RNAseq and microRNAs.\n
LOCATION:Large  Seminar Room\, 1st Floor\, Institute of Pub
 lic Health\, University Forvie Site\, Robinson Way
 \, Cambridge
CONTACT:Alison Quenault
END:VEVENT
END:VCALENDAR