A0505
Title: Prediction of trajectory for variable-domain functional data
Authors: Hidetoshi Matsui - Shiga University (Japan) [presenting]
Yoshikazu Terada - Osaka University; RIKEN (Japan)
Abstract: Functional data analysis (FDA) is one of the most useful methods for analyzing longitudinal data and has been widely used in various fields such as medicine and engineering. Standard FDA methods focus on functional data whose domains are identical for each individual. In contrast, the data is considered where the endpoints of functions differ for each individual. The problem of predicting the trajectory and endpoint of a function observed is approached only up to a specific time point under the condition that a set of other functions is completely observed. The idea of variable-domain functional data and dynamic prediction is applied to achieve this. The analysis of real data demonstrates the effectiveness of the proposed method.
