Nested Bayesian Asymptotics Approach to Big Data Challenge
Gyuhyeong Goh, Ph.D
Department of Statistics, Kansas State University
The biggest subject in recent statistical research is Big Data. Due to the massive sample size and high dimensionality of Big Data, we often face computational and statistical challenges. In this talk, we mainly focus on intractable problems when the sample size is extremely large in a general regression setup. In this case, a major issue is that we are unable to analyze the entire data set routinely due to the limited capacity of personal computer as well as of statistical software. To overcome the obstacle, we introduce a new Bayesian method based on Bernstein-von Mises theorem and Split-and-Conquer technique. The proposed method will be exemplified through intensive simulation studies. This is a joint work with Dr. Wei-Wen Hsu at Department of Statistics, Kansas State University.