摘要
随着信息技术的发展,多变量半连续数据出现在越来越多的研究领域中,其主要特征是多变量数据中的每一个变量都含有过多的零值.然而相较于单变量半连续数据,目前却还未有学者关注于多变量半连续数据的假设检验问题.因此,文章主要研究多变量半连续数据的两总体假设检验问题.针对多变量半连续数据,文章构建一种多元Bernoulli-Normal模型,并提出一种基于多元复合原假设的似然比检验方法.由数据模拟结果表明,相较于经典的Hotelling's T^(2)检验方法,似然比方法具有较低的第Ⅰ类错误率和较高的检验功效.此外,将此方法应用到饮食摄入量CHEF实例数据中,结果表明所提出的方法能够对干预措施的有效性进行评估.
With the development of information technology,multivariate semicontinuous data appears in more and more research fields,whose feature is that each variable in multivariate data contains too many zero values.However,compared with semicontinuous data,no scholars pay attention to the problem of hypothesis testing of multivariate semicontinuous data.Therefore,this paper mainly studies the problem of two-population hypothesis testing for multivariate semicontinuous data.For multivariate semicontinuous data,a multivariate Bernoulli-Normal model is constructed and a likelihood ratio testing method based on multivariate compound null hypothesis is proposed.The data simulation results show that,compared with the classical Hotelling’s T^(2)test,the likelihood ratio method has a lower type Ⅰ error rates and a higher test power.In addition,this method is applied to the dietary intake CHEF data,and the results show that the proposed method could be used to evaluate the effectiveness of the intervention.
作者
鲁亚会
刘爱义
江涛
LU Yahui;LIU Aiyi;JIANG Tao(School of Economics and Management,Zhejiang University of Science and Technology,Hangzhou 310023;National Institutes of Health,USA,Bethesda MD 20817;School of Statistics and Mathematics,Zhejiang Gongshang University,Hangzhou 310018;Hangzhou College of Commerce,Zhejiang Gongshang University,Tonglu 311599)
出处
《系统科学与数学》
CSCD
北大核心
2021年第11期3254-3266,共13页
Journal of Systems Science and Mathematical Sciences
基金
国家自然科学基金(11971433)
浙江科技学院科研启动基金(F701107L04)资助课题。
