Unsupervised classiﬁcation for multivariate functional data
Velasco Mendoza, Javier Antonio
Magíster en Matemáticas Aplicadas
MetadataShow full item record
In this paper, we propose a new method to classify functional data but in the multivariate case. This new technique is based on some order and centrality measures for the functional framework. Although our methodology works well in the general case, in this work each record is composed of two functional variables and the functional sample is composed of several records. We design a statistical tool for segmenting that sample in groups with similar characteristics. We test our methodology with real and simulated data and we highlight that this new method introduced here, work better than those techniques already introduced in the literature.