Title: kmlShape method to cluster longitudinal data
Authors: Soonsun Kwon - Ajou University (Korea, South) [presenting]
Abstract: Longitudinal data arise where a response is observed on each subject repeatedly over the time. One possibility for the analysis of longitudinal data is to cluster them. The majority of clustering methods group together subject that have close trajectories at given time points. The methods group trajectories that are locally close, but not necessarily those that have similar shapes. In contrast, the kmlshape method is a longitudinal data partitioning algorithm based on the shapes of the trajectories, rather than on classical distances. We suggest the kmlShape method for high-dimensional datasets, and apply it to thyroidectomy cancer patients.