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:Average derivative projection pursuit regression -
Frédéric Ferraty\, Toulouse
DTSTART;TZID=Europe/London:20141114T160000
DTEND;TZID=Europe/London:20141114T170000
UID:TALK54673AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/54673
DESCRIPTION:In this talk one proposes a general framework for
projection Pursuit Regression (PPR) that can deal
with an explanatory random function X = {X(t)\; t
∈ I} and a scalar response Y. In doing so we intro
duce what we call the average derivative operator
of the regression. We show that the unknown direct
ions involved in the functional PPR can be viewed
as the eigenfunctions of the average derivative op
erator. This new characterization of the direction
al parameters greatly simplifies the estimation pr
oblem\, offers new interpretation of the direction
al parameters in PPR and solves identifiability is
sues inherent in such models.\n\nMotivated by rece
nt works focusing on the nonparametric estimation
of the directional derivatives of a regression ope
rator\, a nonparametric estimator of the average d
erivative operator is constructed. Asymptotic prop
erties of the introduced estimators are studied an
d numerical examples are used to illustrate the me
thod and to assess finite sample performance.
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberf
orce Road\, Cambridge
CONTACT:
END:VEVENT
END:VCALENDAR