A latent-class mixture model is proposed for the unsupervised classification of incomplete multivariate data with mixed linear and circular components. The model allows for nonignorable missing values and integrates circular and normal densities to capture the association between toroidal clusters of circular observations and elliptical clusters of linear observations. Maximum likelihood estimation of the model is facilitated by an EM algorithm that treats unknown class membership and missing values as different sources of incomplete information. The model is exploited on incomplete time series of wind speed and direction and wave height and direction to identify a number of sea regimes.
Lagona, F., Picone, M. (2013). Classification of multivariate linear-circular data with nonignorable missing values. In GRIGOLETTO M, LISI F, PETRONE S (a cura di), Complex models and computational methods in statistics (pp. 161-173). SPRINGER [10.1007/978-88-470-2871-5_13].
Classification of multivariate linear-circular data with nonignorable missing values
LAGONA, Francesco;
2013-01-01
Abstract
A latent-class mixture model is proposed for the unsupervised classification of incomplete multivariate data with mixed linear and circular components. The model allows for nonignorable missing values and integrates circular and normal densities to capture the association between toroidal clusters of circular observations and elliptical clusters of linear observations. Maximum likelihood estimation of the model is facilitated by an EM algorithm that treats unknown class membership and missing values as different sources of incomplete information. The model is exploited on incomplete time series of wind speed and direction and wave height and direction to identify a number of sea regimes.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
