Abstract

This paper is concerned with the simulation of an extension of the Mallows-Bradley-Terry ranking probability model by the acceptance-rejection method. A Monte Carlo Markov Chain (MCMC) algorithm for the simulation of the model has been already proposed when the number q of items to be ranked is large, say more than 7. However, in most real life situations the number q of items to be ranked does not exceed 10, e.g., psycho physics, food testing, etc. Therefore, the proposed tool relies on appropriate choice of the constant and instrumental distribution by means of the well-known acceptance-rejection method to generate samples from the target distribution.

摘要

本文关注的是通过接受-拒绝方法模拟 Mallows-Bradley-Terry 排序概率模型的扩展。当要排序的项目数q很大时，例如超过 7，已经提出了一种用于模型模拟的蒙特卡洛马尔可夫链 (MCMC) 算法。然而，在大多数现实生活中，要排序的项目数q ranked does not exceed 10, eg, psycho physics, food testing, etc. Therefore, the proposed tool relies on appropriate choice of the constant and instrumental distribution by means of the well-known acceptance-rejection method to generate samples from the target distribution.