Hidden Markov Models for Time Series: An Introduction Using R

摘要

Hidden Markov Models (HMMs) have become increasingly popular in recent years. The introduction of a hidden state gives plenty of flexibility for handling a variety of complex real-world time series, and the Markovian structure of the model remains simple enough to allow interpretation and efficient computational procedures.Despite their popularity there exist few books on HMMs. In particular, there are none at an introductory level that are accessible to a wide audience, as this book is. The previous book by the same authors, Hidden Markov and Other Models for Discrete-Valued Time Series (MacDonald & Zucchini 1997), was a very good and popular introduction to HMMs. Compared with the previous book, the authors have chosen to remove the first part on the 'other models' and focus exclusively on HMMs for a wider range of time series, not only the ones with discrete values. In addition, the authors have made the models more accessible by illustrating how the open-source computing environment R can be used to perform various computations related to HMMs. They do not supply a ready-to-use R package but give code for specific HMMs. Readers are encouraged to develop their own code through the exercises that are provided at the end of each chapter. This is a legitimate choice, as many applications require the adaptation of standard cases that are covered by existing R packages.