B1196
Title: Clustering fragmentary functional data
Authors: Yuguang Fan - University of Melbourne (Australia) [presenting]
Aurore Delaigle - University of Melbourne (Australia)
Abstract: We study curves that are observed only on short and irregular intervals. Such fragmentary functional data is often collected in practice but it poses several difficulties when we try to apply the usual methods designed for curves defined on the entire interval. Under these circumstances, we propose a method to cluster such fragments addressing the difficulties in lateral connection and design an algorithm to optimize the cluster performance.
