B1116
Title: Covariance matrix estimation for functional and longitudinal data
Authors: Uche Mbaka - University College Dublin (Ireland) [presenting]
Michelle Carey - Univerity College Dublin (Ireland)
Abstract: Covariance estimation plays a critical role in various models; however, many existing techniques often fail to provide positive definite estimates, preventing the direct computation of the precision matrix, a vital component in numerous applications. To address the challenge, a novel method is proposed for estimating the lower triangular matrix of the Cholesky factorization of a covariance matrix of functional data using Splines. Subsequently, the estimated lower triangular matrix is used to recover the full covariance matrix. The new approach is comparable to other existing methods in terms of accuracy while yielding an almost always positive definite estimate.
