Read more
Sampling from the posterior distribution and computing posterior quanti ties of interest using Markov chain Monte Carlo (MCMC) samples are two major challenges involved in advanced Bayesian computation. This book examines each of these issues in detail and focuses heavily on comput ing various posterior quantities of interest from a given MCMC sample. Several topics are addressed, including techniques for MCMC sampling, Monte Carlo (MC) methods for estimation of posterior summaries, improv ing simulation accuracy, marginal posterior density estimation, estimation of normalizing constants, constrained parameter problems, Highest Poste rior Density (HPD) interval calculations, computation of posterior modes, and posterior computations for proportional hazards models and Dirichlet process models. Also extensive discussion is given for computations in volving model comparisons, including both nested and nonnested models. Marginal likelihood methods, ratios of normalizing constants, Bayes fac tors, the Savage-Dickey density ratio, Stochastic Search Variable Selection (SSVS), Bayesian Model Averaging (BMA), the reverse jump algorithm, and model adequacy using predictive and latent residual approaches are also discussed. The book presents an equal mixture of theory and real applications.
List of contents
1 Introduction.- 1.1 Aims.- 1.2 Outline.- 1.3 Motivating Examples.- 1.4 The Bayesian Paradigm.- Exercises.- 2 Markov Chain Monte Carlo Sampling.- 2.1 Gibbs Sampler.- 2.2 Metropolis-Hastings Algorithm.- 2.3 Hit-and-Run Algorithm.- 2.4 Multiple-Try Metropolis Algorithm.- 2.5 Grouping, Collapsing, and Reparameterizations.- 2.6 Acceleration Algorithms for MCMC Sampling.- 2.7 Dynamic Weighting Algorithm.- 2.8 Toward "Black-Box" Sampling.- 2.9 Convergence Diagnostics.- Exercises.- 3 Basic Monte Carlo Methods for Estimating Posterior Quantities.- 3.1 Posterior Quantities.- 3.2 Basic Monte Carlo Methods.- 3.3 Simulation Standard Error Estimation.- 3.4 Improving Monte Carlo Estimates.- 3.5 Controlling Simulation Errors.- Exercises.- 4 Estimating Marginal Posterior Densities.- 4.1 Marginal Posterior Densities.- 4.2 Kernel Methods.- 4.3 IWMDE Methods.- 4.4 Illustrative Examples.- 4.5 Performance Study Using the Kullback-Leibler Divergence.- Exercises.- 5 Estimating Ratios of Normalizing Constants.- 5.1 Introduction.- 5.2 Importance Sampling.- 5.3 Bridge Sampling.- 5.4 Path Sampling.- 5.5 Ratio Importance Sampling.- 5.6 A Theoretical Illustration.- 5.7 Computing Simulation Standard Errors.- 5.8 Extensions to Densities with Different Dimensions.- 5.9 Estimation of Normalizing Constants After Transformation.- 5.10 Other Methods.- 5.11 An Application of Weighted Monte Carlo Estimators.- 5.12 Discussion.- Exercises.- 6 Monte Carlo Methods for Constrained Parameter Problems.- 6.1 Constrained Parameter Problems.- 6.2 Posterior Moments and Marginal Posterior Densities.- 6.3 Computing Normalizing Constants for Bayesian Estimation.- 6.4 Applications.- 6.5 Discussion.- Exercises.- 7 Computing Bayesian Credible and HPD Intervals.- 7.1 Bayesian Credible and HPD Intervals.- 7.2 EstimatingBayesian Credible Intervals.- 7.3 Estimating Bayesian HPD Intervals.- 7.4 Extension to the Constrained Parameter Problems.- 7.5 Numerical Illustration.- 7.6 Discussion.- Exercises.- 8 Bayesian Approaches for Comparing Nonnested Models.- 8.1 Marginal Likelihood Approaches.- 8.2 Scale Mixtures of Multivariate Normal Link Models.- 8.3 "Super-Model" or "Sub-Model" Approaches.- 8.4 Criterion-Based Methods.- 9 Bayesian Variable Selection.- 9.1 Variable Selection for Logistic Regression Models.- 9.2 Variable Selection for Time Series Count Data Models.- 9.3 Stochastic Search Variable Selection.- 9.4 Bayesian Model Averaging.- 9.5 Reversible Jump MCMC Algorithm for Variable Selection.- Exercises.- 10 Other Topics.- 10.1 Bayesian Model Adequacy.- 10.2 Computing Posterior Modes.- 10.3 Bayesian Computation for Proportional Hazards Models.- 10.4 Posterior Sampling for Mixture of Dirichlet Process Models.- Exercises.- References.- Author Index.
Summary
Bayesian statistics is one of the active research areas in statistics. This book provides the theoretical background behind the important development, Markov chain Monte Carlos methods.
Report
"This book combines the theory topics with good computer and application examples from the field of food science, agriculture, cancer and others. The volume will provide an excellent research resource for statisticians with an interest in computer intensive methods for modelling with different sorts of prior information."
A.V. Tsukanov in "Short Book Reviews", Vol. 20/3, December 2000
