Online Program Home
My Program

Abstract Details

Activity Number: 30 - SPEED: Statistics and Econometrics
Type: Contributed
Date/Time: Sunday, July 29, 2018 : 2:00 PM to 3:50 PM
Sponsor: Business and Economic Statistics Section
Abstract #329934 Presentation
Title: Two-Sample Test for Covariance Operators with Incompletely Observed Functional Data
Author(s): Lihan Yan* and Tao Zhang and Zhaohai Li
Companies: FDA and Guangxi University of Science and Technology and George Washington University
Keywords: Functional data analysis; Covariance operators; significance test; Incompletely observed functional data

In this presentation, we present a four-step procedure to test the equality of two covariance operators in functional data samples. This method covers the situation where partial functional data are fully observed on dense lattice and other partial are incompletely observed, as well as the case where the recordings of all of curves are available on a regular and dense grid. The procedure leads to a global testing statistic. In the procedure, we obtain projected covariance operators based on an initial functional principal component analysis and then form a global test statistic based on the projected covariance operators and their corresponding variance estimates. The asymptotic properties of the test statistic are presented. Simulation results demonstrate that the test controls Type I error rates at a reasonable level and may provide superior power results compared with the existing methods. We show a real life application of the proposed test involving air pollution data to test whether there are conjectured differences of volatility in air pollution between working days and non-working days, and between seasons.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program