Oracally Efficient Spline Smoothing of Functional Coefficient Regression Models with Simultaneous Confidence Band

• Activity Number: 415
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Weixin Cai*+ and Prabir Burman and Joshua Patrick
• Companies: University of California, Davis and University of California, Davis and University of California, Davis
• Keywords: Nonlinear time series ; confidence band ; oracle efficiency

Abstract

This poster examines the confidence band for fitting Functional Coefficient Autoregressive (FCAR) model by implementing spline-backfitting technique. The technique is both computationally expedient for analyzing high dimensional time series data, and theoretically reliable as the estimators are oracally efficient. The method will also provide asymptotic results that allow for confidence bands for the estimates of coefficient functions. The feasibility of both spline-backfitted kernel and spline-backfitted spline approach will be investigated. The new method will be illustrated with stimulation results and "real-world" data examples.