A0212
Title: Bayesian model selection for high-dimensional data integration
Authors: Yuehua Wu - York University (Canada) [presenting]
Abstract: Data integration problems are considered when correlated data are collected from multiple platforms. Each platform has linear relationships between the responses and a collection of predictors. The linear models are extended to include random errors from a much wider family of sub-Gaussian and sub-exponential distributions. The goal is to select important predictors across multiple platforms, where the number of predictors and the number of observations both increase to infinity. The marginal densities of the responses obtained from different platforms are combined to form a composite likelihood and propose a model selection criterion based on Bayesian composite posterior probabilities. Under some regularity conditions, it is proven that the model selection criterion is consistent with divergent true model sizes. A Monte Carlo Markov Chain algorithm is implemented for the model selection approach. Simulation results and a real data example are further presented.
