A Novel Nonparametric Clustering Method for Longitudinal Data

• Activity Number: 79 - Functional Data Analysis: Methods and Applications
• Type: Contributed
• Date/Time: Sunday, July 28, 2019 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Junyi Zhou* and Ying Zhang
• Companies: Indiana University and University of Nebraska Medical Center
• Keywords: clustering; unbalanced functional data; spline regression; subgroups

Abstract

We propose an easy-to-compute spline-based non-parametric clustering method for longitudinal data. The approach can be readily applied to various types of functional data especially when data are sparsely sampled, and/or irregular spaced. In addition, this approach can also handle the situation with unbalanced functional data, which is commonly occurred in applications, however, has not been seriously taken into account in the literatures. We conduct an extensive simulation study to demonstrate that the proposed method works well and outperforms the existing KmL approach that has an available package with R. Finally, we apply the proposed method to data from a 12-year observational cohort study of premanifest Huntington disease subjects to identify the phenotypes of disease progression.