Resumen de Regularisation in hidden markov models with an application to football data
We propose a modelling framework for dealing with a large amount of covariates in hidden Markov models (HMMs) by considering a LASSO penalty.
This modelling framework is, for example, useful in sports for analysing a potential hot hand e ect, as several existing studies on the hot hand consider HMMs.
However, with most studies analysing data from basketball or baseball, there are several confounding factors which have to be taken into account, leading to a potential large number of covariates. Hence, in those settings regularisation methods are suitable to allow for implicit variable selection. As a case study we investigate a potential \hot shoe" e ect among penalty-takers.
