Research Seminar
Visualization of distributions of covariance matricesTomoki TokudaKU Leuven | |
| Abstract: | Covariance matrices play an important role in multivariate statistics, as they include information on the variability and associations of the variables under study. In Bayesian data analysis, often a prior distribution (such as an inverse Wishart distribution) has to be introduced for covariance matrices. However, analytical results on the properties of such prior distributions are in general small in number; moreover, the visualization of the distributions in question implies a major challenge because of the high-dimensionality of their support. As a way out, in this talk, a novel approach will be proposed to visualize distributions of covariance matrices in terms of a four-layered graphical representation. Furthermore, we will show how an application of the approach to a posterior distribution of a covariance matrix may provide insights that are helpful for the specification of tuning parameters in the corresponding (conjugate) prior distribution. |
| Date: | Tue Feb 15, 12:15 pm - 1:15 pm |
| Place: | room 02.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |