Bayesian Clustering of Categorical Time Series: An Approach Using Finite Mixtures of Markov Chain Models

In many areas of applied statistics like economics or finance it is often desirable to find groups of similar time series in a set or panel of time series. Therefore, clustering techniques are required to determine subsets of similar time series. While distance-based clustering methods cannot easily be extended to time series data, model-based clustering based on finite mixture models extends to time series data in quite a natural way. The author Christoph Pamminger proposes and discusses two approaches for model-based clustering methods specifically designed for categorical time series data and presents an application of these methods to a panel of Austrian wage mobility data. The aim of this research work was to investigate Austrian wage mobility and to search for groups of employees with similar wage mobility behaviour. The results show an interesting segmentation of the Austrian labour market. This book is suitable and will be interesting for all statisticians, researchers in related fields and any user of statistical methods, especially for those who are concerned with time series data and/or clustering of data.

Christoph Pamminger, Mag. Dr.: Studies of Statistics at Johannes Kepler University Linz, Austria. Research Assistant at the Department of Applied Statistics, JKU Linz (www.ifas.jku.at).