Completion of Partially Observed Multivariate Functional Data

Abstract

Functional data often present missing values. These pose challenges when performing functional principal components analysis, since it is not possible to compute the scores from the data. The existing method by Kraus (2015) allows to perform functional data completion in a univariate setting. In this paper, we propose an extension to the multivariate setting and show how statistical dependences between the functional curves aid in the imputation of missing data.