Unsupervised classification of marine data is helpful to identify relevant sea regimes, i.e. specific shapes that the distribution of wind and wave data takes under latent environmental conditions. We cluster multivariate marine data by estimating a multivariate hidden Markov model that integrates multivariate von Mises and normal densities. Taking this approach, we obtain a classification that accounts for the mixed (linear and circular) support of the observations, the temporal autocorrelation of the data and the occurrence of missing values.
Bencivenga, M., Lagona, F., Maruotti, A., Nardone, G., Picone, M. (2016). Unsupervised classification of multivariate time series data for the identification of sea regimes.. In GIOMMI A., ALLEVA G. (a cura di), Topics in Theoretical and Applied Statistics (pp. 61-71). Springer [10.1007/978-3-319-27274-0_6].
Unsupervised classification of multivariate time series data for the identification of sea regimes.
LAGONA, Francesco;
2016-01-01
Abstract
Unsupervised classification of marine data is helpful to identify relevant sea regimes, i.e. specific shapes that the distribution of wind and wave data takes under latent environmental conditions. We cluster multivariate marine data by estimating a multivariate hidden Markov model that integrates multivariate von Mises and normal densities. Taking this approach, we obtain a classification that accounts for the mixed (linear and circular) support of the observations, the temporal autocorrelation of the data and the occurrence of missing values.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
