HARK: A New Approach for Regression with Functional Predictors

• Activity Number: 661
• Type: Topic Contributed
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Dawn B. Woodard*+
• Companies: Cornell University
• Keywords: functional data analysis ; regression ; Bayesian ; Levy process ; modularization ; kernel mixture

Abstract

We propose a new method for regression using functional predictors. We use a parsimonious functional data model with parameters that have intuitive scientific interpretation. Our approach is designed for data that include spikes, dips, plateaus, or other features whose location, size, and shape varies across subjects. The functional parameters are then used to predict outcomes. We propose full Bayesian inference of the joint functional regression and exposure models, and give a method for efficient computation. We contrast our approach with existing methods for regression with functional predictors, and show that our method can have much higher power for data that include features occurring at varying locations. We also apply our methodology to a large and complex dataset from the Sleep Heart Health Study, in order to better understand the effect of sleep patterns on health outcomes.