Sequential Gibbs Sampling Algorithm for Cognitive Prognosis Fashions with Many Attributes
Authors: Juntao Wang, Ningzhong Shi, Xue Zhang, Gongjun Xu
Summary: Cognitive prognosis fashions (CDMs) are helpful statistical instruments to supply wealthy info related for intervention and studying. As a preferred method to estimate and make inference of CDMs, the Markov chain Monte Carlo (MCMC) algorithm is broadly utilized in apply. Nevertheless, when the variety of attributes, Okay, is giant, the prevailing MCMC algorithm could turn into time-consuming, attributable to the truth that O(2K) calculations are normally wanted within the technique of MCMC sampling to get the conditional distribution for every attribute profile. To beat this computational problem, motivated by Culpepper and Hudson (2018), we suggest a computationally environment friendly sequential Gibbs sampling methodology, which wants O(Okay) calculations to pattern every attribute profile. We use simulation and actual knowledge examples to point out the nice finite-sample efficiency of the proposed sequential Gibbs sampling, and its benefit over current strategies
