Publication Results

Code: 42/2019
Title: hmmhdd Package: Hidden Markov Model for High Dimensional Data
Date: Saturday 9th November 2019
Author(s) : Martino, A.; Guatteri, G.; Paganoni, A.m.
Download link:
Abstract: The R package "hmmhdd" provides some tools to study times series and longitudinal datasets. In particular, the package is based on Hidden Markov Models, i.e. it considers an underlying structure defined by a Markov Model with non-observable states generating a certain type of data, in the multivariate or functional framework. In the former setting, a Gaussian copula models the correlation structure between the components of the observations while, in the latter setting, the data are multivariate functional data and the methods are based on distances between curves. The package is able to estimate all the parameters corresponding to the states of the underlying Markov model, while also computing the optimal state sequence and providing some further helpful tools.