Intrinsic Riemannian functional data analysis (Q2284383): Difference between revisions

Although traditionally function data take values in a vector space, more data of nonlinear structure arise and should be handled in a nonlinear space. The paper under review works with the underlying Riemannian manifold without assuming an ambient space surrounding it or an isometric embedding into a Euclidean space. This approach is regarded as completely intrinsic. The key building block is a new concept of tensor Hilbert space along a curve on the manifold. The authors develop intrinsic Riemannian functional principal component analysis (iRFPCA) and intrinsic Riemannian functional linear regression (iRFLR). Specifically, estimation procedures of intrinsic eigenstructure are provided and their asymptotic properties are investigated. For iRFLR, a Riemannian functional linear regression model is studied, where a scalar response intrinsically and linearly depends on a Riemannian functional predictor through a Riemannian slope function. The paper presents an iRFPCA-based estimator and a Tikhonov estimator for the Riemannian slope function and explores their asymptotic properties. Numerical performance is illustrated by simulations and application to Human Connectome Project analyzing functional connectivity and behavioral data.