Bayesian Statistics Seminar
North Carolina State University
presents
Dr. Subhashis Ghosal
North Carolina State University
"Convergence rate of Bayesian variable selection in GLMs with number of terms K>>n under sparseness"
ABSTRACT
Consider n observations from a generalized linear model with K>>n variables, but the true values of the coefficients satisfy an implicit sparseness condition. Suppose variables are a priori independently included in the model with an appropriate probability. I will present a recent result from the literature which investigates convergence rate to the true distribution. It is also shown by an example that some sparseness condition is necessary for such a result to hold.
Tuesday, October, 17, 2006
4:00 - 5:00 pm
208 Patterson Hall