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:Statistics
SUMMARY:Simple Methods to Improve MCMC Efficiency in Rando
m Effect Models - Bill Brown (University of Bristo
l)
DTSTART;TZID=Europe/London:20081205T160000
DTEND;TZID=Europe/London:20081205T170000
UID:TALK14326AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/14326
DESCRIPTION:MCMC methods have continued to grow in popularity
as their flexibility\nin terms of the vast number
of models they can fit is realised. The\nfamily of
MCMC algorithms is large and many applied researc
hers exposure\nto MCMC methods is through their us
e of the default estimation methods\nprovided in s
oftware packages such as WinBUGS or MLwiN. Althoug
h these\npackages often try to optimize the steps
of the algorithm they use to\nfit particular model
s they can still produce algorithms that result in
\npoorly mixing chains. Many statistical methodolo
gists produce model\nspecific methods to improve m
ixing and create efficient MCMC algorithms\,\nbut
for this methodology to impact on the applied comm
unity it needs to\nbe implemented in available sof
tware. One particular way to improve the\nefficien
cy of an MCMC algorithm is through model re-parame
terisation.\nSome reparameterisation methods can b
e easily implemented by\nmodifications to the mode
l code input into WinBUGS or via some\nforthcoming
developments in MLwiN.\nIn this talk we describe
briefly three such reparameterisation\ntechniques\
, hierarchical centering (Gelfand et al. 1995)\, p
arameter\nexpansion (Liu et al. 1998) and orthogon
alisation of the fixed\npredictors (Browne et al.
submitted) which can be easily implemented in WinB
UGS. We will show\nhow these methods perform on a
selection of random effect models applied to\nexa
mples from ecology\, veterinary epidemiology and d
emography.\n\n
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0W
B
CONTACT:
END:VEVENT
END:VCALENDAR