 |
Center for Nonlinear Studies |
In the past decades hidden semi-Markov models (HSMMs) have become increasingly popular. They provide flexible, general-purpose models for univariate and multivariate time series. The observed component may include categorical, circular-valued, and many other observation types. Moreover, the extension to a panel data setting is straightforward. Appealing features of HSMMs include their versatility and mathematical tractability, which have led to applications in many fields (e.g. engineering, finance/econometrics, and environmental studies). The main objective of this talk is to provide an introduction to the basic underlying concepts of HSMMs, and their estimation. Their practical relevance is illustrated by means of an application to a longitudinal data set of EEG measurements, collected over an unsupervised period of 24 hours.
Host: Thomas Leitner