．單位首頁
．系所簡介
．系主任
．強力師資
．研究所
．課程概況
．各類學程
．期刊論文
．招生資訊
．演講專區
．榮譽榜
．系學會
．系友天地
．就業訊息
96年10月19日(五) 2:00 ~3:00 P.M.
演講者姓名： 李百靈教授
演講者服務單位： Department of Statistics, TamkangUniversity
K-Centers Functional Clustering of Longitudinal Data
Abstract
We propose a novel clustering method, k-centers functional clustering (FC) algorithm, for longitudinal data. Under the assumption that each cluster’s center consists of the stochastic structures including the mean and modes of variation, the k-centers FC is iteratively implemented by reclassifying each subject to the best predicted cluster membership. The reclassification procedure is based on fitting curves by a nonparametric random effect model of the truncated Karhunen-Loeve expansion. We show that the proposed clustering algorithm can improve the cluster quality under certain conditions. In addition, the stochastic structures for the clusters are naturally identified through the k-centers FC method. The performance of our proposed method is demonstrated by simulation studies and examples.
