New Statistical Diagnostics Based On Posterior Samples In Spatio-temporal Independent Regression Models

In this paper, we consider some new statistical diagnostics in spatio-temporal independent regression models. We raise three new Bayesian statistical diagnostic-s based on posterior samples:Kulback-Leibler distance of kernel density, Cook distance, volume ratio of confidence region and distance of confidence region.In the first chapter, we review the research work of statistical inference, and then we list some important preliminary knowledge.In the second chapter, we design the three new Bayesian statistical diagnostics.In the third chapter, by using the technique of the stochastic simulation in linear regression models model, we check the effectiveness of the statistical diagnostics.In the forth chapter, by using the technique of the stochastic simulation in spatio-temporal independent regression models, we check the effectiveness of the statistical diagnostics.In the last chapter, in comparison of the three new statistical diagnostics and the traditional Kulback-Leibler distance, we make a summary and propose some relevant research work in the future.