ARTICLE
ABSTRACT
In classical data analysis, objects are characterized by many features observed at one point of time. We would like to present them graphically, to see their configuration, eliminate outlying observations, observe relationships between them or to classify them. In recent years methods for representing data by functions have received much attention. In this paper we discuss a new method of constructing principal components for multivariate functional data. We illustrate our method with data from environmental studies
KEYWORDS
multivariate functional data, functional data analysis, principal component analysis, multivariate principal component analysis.
REFERENCES
BESSE, P., (1979). Etude descriptive d’un processus. Ph.D. thesis, Université Paul Sabatier.
FISHER, R. A., (1936). The use of multiple measurements in taxonomic problem, Annals of Eugenics 7, 179–188.
GÓRECKI, T., KRZYŚKO, M., (2012). Functional Principal Components Analysis, Data analysis methods and its applications, C.H. Beck 71–87.
GÓRECKI, T., KRZYŚKO, M., WASZAK, Ł., (2014). Functional Discriminant Coordinates, Communication in Statistics – Theory and Methods 43(5), 1013–1025.
JACQUES, J., PREDA, C., (2014). Model-based clustering for multivariate functional data, Computational Statistics & Data Analysis 71, 92–106.
JOLLIFFE, I. T., (2002). Principal Component Analysis, Second Edition, Springer.
KRZYŚKO, M., WASZAK, Ł., (2013). Canonical correlation analysis for functional data, Biometrical Letters 50(2), 95–105.
RAMSAY, J. O., SILVERMAN, B. W., (2005). Functional Data Analysis, Second Edition, Springer.
SAPORTA, G., (1981). Methodes exploratoires d’analyse de données temporelles. Cahiers du Buro. Ph.D. thesis.
SHMUELI, G., (2010). To explain or to predict? Statistical Science 25(3), 289–310.
