Crisp and fuzzy k-means clustering algorithms for multivariate functional data (Q2271688)

The authors consider data which are a set of vector-valued functions on a segment of the real line. Generalizations of MacQueen's crisp and Ruspini's fuzzy \(k\)-means clustering algorithms for such data are proposed. Examples of artificial data and a growth study data analysis are presented. B-spline smoothing is used in the growth data example to transform the original discrete observations into functional data.