Tuesday, November 9, 3.15 pm, 2021. Seminar in Statistics.

Hypothesis Testing in Multivariate Normal Models with Block Circular Covariance Structures
Yuli Liang, Örebro University School of Business
Abstract: In this article, we address the problem of simultaneous testing hypothesis about mean and covariance matrix for repeated measures data when both the mean vector and covariance matrix are patterned. In particular, tests about the mean vector under block circular and doubly exchangeable covariance structures have been considered. The null distributions are established for the corresponding likelihood ratio test statistics and expressions for the exact or near-exact probability density and cumulative distribution functions are obtained. The application of the results is illustrated by both a simulation study and a real-life data example.
Location: Online via Zoom. Please email Krzysztof Bartoszek for invitation to Zoom meeting.

Tuesday, December 7, 3.15 pm, 2021. Seminar in Statistics.

Introduction to layeranalyzer, a process-based R package for time series analysis
Trond Reitan, Natural History Museum, University of Oslo, Norway
Abstract:By employing statistical continuous time process models, one can perform time series analysis even when the measurements are distributed haphazardly in time (as they often are in the field of paleontology). Even when the measurements are equidistant in time, one may want a model that is continuous in time. Continuous time processes can be cumbersome to calculate likelihoods for, which affects the ability to perform Bayesian or ML-based parameter estimation. We opted to restrict ourselves to a wide suite of linear stochastic differential equations, so that we can model in continuous time while still calculating likelihoods efficiently. The R package 'layeranalyzer', which will be presented here, uses this framework to perform time series analysis, including parameter estimation, model testing, process inference and interpolation/extrapolation. When it comes to connections between processes, it also distinguishes between (Granger) causal and correlative connections. The option of causal connections allows for exploring so-called "hidden layers" (thus the name of the package), that is unmeasured processes that affects measured processes. I will describe some technical challenges with this framework, before delving into a couple of usage examples.
Location: Online via Zoom. Please email Krzysztof Bartoszek for invitation to Zoom meeting.