BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Simultaneous multiple change-point and factor analysis for high-di
 mensional time series - Haeran Cho (Bristol)
DTSTART:20170210T160000Z
DTEND:20170210T170000Z
UID:TALK70611@talks.cam.ac.uk
CONTACT:Quentin Berthet
DESCRIPTION:We propose the first comprehensive treatment of high-dimension
 al time series factor models with multiple change-points in their second-o
 rder structure. We operate under the most flexible definition of piecewise
  stationarity\, and estimate the number and locations of change-points con
 sistently as well as identifying whether they originate in the common or i
 diosyncratic components. Through the use of wavelets\, we transform the pr
 oblem of change-point detection in the second-order structure of a high-di
 mensional time series\, into the (relatively easier) problem of change-poi
 nt detection in the means of high-dimensional panel data. Our methodology 
 circumvents the difficult issue of the accurate estimation of the true num
 ber of factors by adopting a screening procedure. In extensive simulation 
 studies\, we show that factor analysis prior to change-point detection imp
 roves the detectability of change-points\, and identify and describe an in
 teresting ‘spillover’ effect in which substantial breaks in the idiosy
 ncratic components get\, naturally enough\, identified as change-points in
  the common components\, which prompts us to regard the corresponding chan
 ge-points as also acting as a form of ‘factors’. We introduce a simple
  graphical tool for visualising the piecewise stationary evolution of the 
 factor structure over time. Our methodology is implemented in the R packag
 e factorcpt\, available from CRAN. \n\nJoint work with Matteo Barigozzi an
 d Piotr Fryzlewicz (LSE).
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberforce Road\, Camb
 ridge.
END:VEVENT
END:VCALENDAR