Webinar video https://www.youtube.com/watch?v=bKgbqLIPVgk
Presentation slides
Abstract
Consider the case where several time series are observed, but the variables of interest are ordinal. The goal is to account for both the temporal correlation within each series and the cross-correlation between different series.
In this talk, we will present models designed to address this setting. Modeling multivariate ordinal variables is generally challenging, and extending such models to time series data adds an additional layer of complexity.
We will demonstrate how univariate discrete autoregressive models can be extended to handle this situation, focusing initially on the bivariate case.
We will define a Bivariate Discrete Autoregressive model and discuss in detail its properties and estimation.
We will then show how the multivariate case can be approached using a pairwise likelihood framework, where only the bivariate components need to be specified.
Simulation studies illustrating the performance of the estimation method will be discussed, along with theoretical aspects of the proposed models and applications to real data.
Short Bio
Dimitris Karlis is a Professor in the Department of Statistics at the Athens University of Economics and Business (AUEB). He received his B.Sc. in Statistics from AUEB in 1992 and his Ph.D. in Statistics from the same department in 1999.
He has published more than 115 papers in peer-reviewed statistical journals. His research interests include mixture models, discrete values time series, computational statistics (with emphasis on stochastic algorithms), multivariate count data analysis, statistical modeling for sports data, dependence modeling via copulas among others.
Professor Karlis serves as Associate Editor for Metron and Communications in Statistics (both Theory and Methods and Computation and Simulation). He has also served as Editor-in-Chief of the Journal of Quantitative Analysis in Sports and as Associate Editor for the IMA Journal of Management Mathematics and Stochastic Environmental Research and Risk Assessment. In addition, he was the editor of the Biometrics Bulletin of the International Biometric Society (2013–2015).
He has supervised five Ph.D. students and more than 100 Master’s theses, and he is currently supervising five additional Ph.D. students. Professor Karlis has been invited to present his work at numerous international conferences around the world.
He is a member of the American Statistical Association, an elected member of the International Statistical Institute, a member of the International Association of Statistical Computing, and a former President of the Eastern Mediterranean Region of the International Biometric Society. He is also a member of the Greek Statistical Institute and has participated in several European projects related to statistics, particularly in the field of official statistics.
Information at FEP News
More information at News U. Porto