A review on Clustering Methods for Functional Data

Abstract

Many studies have been done for clustering functional data as considerable functional data are obtained recently. We reviewed overall approaches for clustering functional data proposed so far. Those approaches consist of a nonparametric approach which uses dissimilarity between curves as dissimilarity measure, a filtering and clustering technique which is simple and intuitive and a model-based clustering method which assumes a probability distribution of finite dimensional coefficients estimated from data. Model-based methods are reviewed in detail, particularly. Also, we provided an application to energy data using model-based models for functional data to illustrate model-based methods with specific basis.