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:Simultaneous multiple change-point and factor anal
ysis for high-dimensional time series - Haeran Cho
(Bristol)
DTSTART;TZID=Europe/London:20170210T160000
DTEND;TZID=Europe/London:20170210T170000
UID:TALK70611AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/70611
DESCRIPTION:We propose the first comprehensive treatment of hi
gh-dimensional time series factor models with mult
iple change-points in their second-order structure
. We operate under the most flexible definition of
piecewise stationarity\, and estimate the number
and locations of change-points consistently as wel
l as identifying whether they originate in the com
mon or idiosyncratic components. Through the use o
f wavelets\, we transform the problem of change-po
int detection in the second-order structure of a h
igh-dimensional time series\, into the (relatively
easier) problem of change-point detection in the
means of high-dimensional panel data. Our methodol
ogy circumvents the difficult issue of the accurat
e estimation of the true number of factors by adop
ting a screening procedure. In extensive simulatio
n studies\, we show that factor analysis prior to
change-point detection improves the detectability
of change-points\, and identify and describe an in
teresting ‘spillover’ effect in which substantial
breaks in the idiosyncratic components get\, natur
ally enough\, identified as change-points in the c
ommon components\, which prompts us to regard the
corresponding change-points as also acting as a fo
rm of ‘factors’. We introduce a simple graphical t
ool for visualising the piecewise stationary evolu
tion of the factor structure over time. Our method
ology is implemented in the R package factorcpt\,
available from CRAN. \n\nJoint work with Matteo Ba
rigozzi and Piotr Fryzlewicz (LSE).
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberfo
rce Road\, Cambridge.
CONTACT:Quentin Berthet
END:VEVENT
END:VCALENDAR