A0554
Title: Clustering multivariate functional data: An application of the multivariate epigraph and hypograph indexes
Authors: Belen Pulido Bravo - Universidad Carlos III de Madrid (Spain) [presenting]
Rosa Lillo - Universidad Carlos III de Madrid (Spain)
Alba Franco-Pereira - Universidad Complutense de Madrid (Spain)
Abstract: Dealing with functional data is a challenging problem since it involves working with an infinite-dimension problem. Multivariate functional data adds another layer of complexity by introducing multiple functions that may be related to each other. A major challenge when working with this type of data is how to order it, as there are potentially infinite observations to compare and contrast. There are several proposals in the literature. It is proposed to order multivariate functional data by generalizing the concepts of epigraph and hypograph indexes, which are very useful in the univariate functional data context, is proposed into the multivariate one. After that, clustering multivariate functional data by applying these indexes is proposed to reduce the dimension of the initial dataset. By applying clustering techniques to this final dataset, groups of functions that share similar features can be identified, and insights are gained into the underlying structure of the data. This approach has broad applications in fields such as neuroscience, economics, and environmental sciences, where multivariate functional data are common but challenging to analyze. This new approach has been evaluated against different options available in the literature and applied the methodology to a real data set.
