Comet Calendar Event Details

Statistics Colloquium by Jing Wang
Friday, Nov 8, 2013
(2 p.m. - 3 p.m.)
Location:
JO 4.614
Contact Info:
John Zweck
972-883-6699
Questions? Email me.

 

Jing Wang

Mathematics, Statistics, and Computer Science

University of Illinois at Chicago

Spline Confidence Bands for Functional Derivatives

We develop in this paper a new procedure to construct simultaneous confidence bands for derivatives of mean curves in functional data analysis. The technique involves polynomial splines that provide an approximation to the derivatives of the mean functions, the covariance functions and the associated eigenfunctions. We show that the proposed procedure has desirable statistical properties. In particular, we first show that the proposed estimators of derivatives of the mean curves are semiparametrically efficient. Second, we establish consistency results for derivatives of covariance func- tions and their eigenfunctions. Most importantly, we show that the proposed spline confidence bands are asymptotically efficient as if all random trajectories were observed with no error. Finally, the confidence band procedure is illustrated through numerical simulation studies and a real life example. It is a joint work with Drs. Guanqun Cao, Li Wang, and David Todem.

Sponsored by the Department of Mathematical Sciences

Host:  Qiongxia (Joanne) Song

 



Tagged as Lectures/Seminars

View other events on the Comet Calendar