2016/07/27 by Hang J. Kim, Kim, Hang J., Steven N. MacEachern +3
Computer Science · Mathematics · #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #Markov Chains and Monte Carlo Methods #Methodology (stat.ME) #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.1607.08274
openalex publication_date 2016/07/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Markov chain Monte Carlo samplers produce dependent streams of variates drawn\nfrom the limiting distribution of the Markov chain. With this as motivation, we\nintroduce novel univariate kernel density estimators which are appropriate for\nthe stationary sequences of dependent variates. We modify the asymptotic mean\nintegrated squared error criterion to account for dependence and find that the\nmodified criterion suggests data-driven adjustments to standard bandwidth\nselection methods. Simulation studies show that our proposed methods find\nbandwidths close to the optimal value while standard methods lead to smaller\nbandwidths and hence to undersmoothed density estimates. Empirically, the\nproposed methods have considerably smaller integrated mean squared error than\ndo standard methods.\n