非正态纵向数据随机生成的Monte Carlo模拟方法  

Skewed Longitudinal Data Simulation Based on Monte Carlo Method

作　　者：庄严[1] 杨嘉伟[1] 陈平雁[1] 

机构地区：[1]南方医科大学生物统计学系,510515

出　　处：《中国卫生统计》2015年第3期404-406,共3页Chinese Journal of Health Statistics

基　　金：国家自然科学基金(81402758;81373098);广东省"大学生创新创业训练计划"(1212113040)

摘　　要：目的建立非正态纵向数据的随机序列生成方法。方法根据任意累积分布函数均服从[0,1]的均匀分布原理,通过反函数计算即可将标准多元正态分布随机序列转换为任意分布下的目标随机序列,从而随机生成非正态纵向数据。结果分别对指数分布和二项分布相关序列进行模拟随机数生成,通过对样本的统计分析,均满足预先设定的结果。结论本文提出的非正态纵向数据的随机生成的Monte Carlo模拟方法可产生任意分布下的相关随机序列,不受变量类型限制,方法简单,且具有理论依据,为非正态纵向数据的研究提供了模拟基础。Objective Proposethe random sequence generation method of skewed longitudinal data. Methods Any cumulative distribution function obey standard uniform distribution, according to this principle, we can computethe inverse function, and convertthe random sequence from standard multivariate normal distribution to targeted random sequence from arbitrarily distribution,randomly generateskewed longitudinal data ultimately. Results Respectively, we simulate the exponential distribution and the binomial distribution, and the results have met pre-setthrough the statistical analysis to the samples. Conclusion This paper propose a method based on Monte Carlo simulation which can generatethe random sequencefrom any distributionwithout any limitation to the type of the variates. This method is simple, and it provide a theoretical basis for generating therandomsequence of skewed longitudinal data.

关 键 词：MONTE Carlo 纵向数据 相关序列 随机数 

分 类 号：R195.1[医药卫生—卫生统计学]