A1300
Title: Additive regression for Riemannian functional responses
Authors: Jeong Min Jeon - Ewha Womans University (Korea, South)
Germain Van Bever - Universite de Namur (Belgium) [presenting]
Abstract: Additive regression is explored for a functional response whose values lie on a general Riemannian manifold. Euclidean predictors that may not be directly observable but are estimable are also addressed, such as component scores obtained from dimension reduction. The smooth backfitting method is employed to estimate the additive model, and its asymptotic properties are derived. Additionally, novel dimension reduction techniques are discussed for general predictors in the presence of the Riemannian functional response. The usefulness of the approach is demonstrated through a real data application.
