DEPARTMENT OF STATISTICS
Joint UConn - UMass Statistics Colloquium
The Department of Statistics Cordially Invites You
To a Colloquium
Andrew Bray
Assistant Professor of Statistics
Five College Post-Doctoral Research Associate
University of Massachusetts, Amherst
Approximately Exact Calculations for Linear Mixed Models
Abstract
One of the primary ways that statisticians learn about a linear mixed model in the context of a given data set is to consider the restricted likelihood (RL) function or the joint posterior. Of particular interest are the regions of the parameter space where these functions are high. These regions are traditionally found using general optimizing algorithms that may fail to find the global optimum. In this talk we present a method to optimize the RL function and posterior that is specific to mixed models with two variances. It is a branch and bound algorithm that evaluates the function to an arbitrary degree of precision within a given region of the parameter space, ensuring that no optima are excluded.
DATE: Wednesday, April 15, 2015
TIME: 4:00 p.m.
PLACE: Philip E. Austin Building (AUST) Room 105
Coffee will be served at 3:30 p.m. (AUST 326)
Pizza will be served after the colloquium
For more information, contact: Tracy Burke at tracy.burke@uconn.edu