基于蒙特卡洛方法的假设检验问题探讨  

Discussion on Hypothesis Testing Based on Monte Carlo Method

作　　者：魏艳华[1] 王丙参[1] 邢永忠 Wei Yanhua;Wang Bingcan;Xing Yongzhong(School of Mathematics and Statistics ,Tianshni Normal University,Tianshui Gansu 741001,China;School of Electronic Information and Electrical Engineering,Tianshni Normal University,Tianshui Gansu 741001,China)

机构地区：[1]天水师范学院数学与统计学院,甘肃天水741001 [2]天水师范学院电子信息与电气工程学院,甘肃天水741001

出　　处：《统计与决策》2018年第24期75-78,共4页Statistics & Decision

基　　金：国家自然科学基金资助项目(61104045);天水师范学院中青年教师科研资助项目(TSA1506)

摘　　要：文章利用蒙特卡洛方法产生参考数据去逼近统计量的分布,模拟假设检验的两类错误发生概率与p值。在非平稳序列场合构造一元动态回归模型,探讨虚假回归现象出现的原因与影响因素。对于分布族的卡方拟合优度检验,得出:当样本量较少时,使用渐进分布误差较大,建议使用蒙特卡洛检验方法。This paper uses Monte Carlo method to generate reference data to approximate the distribution of statistics, and simulates two types of error occurrence probability and P value of the hypothesis test. And then the paper constructs a unitary dynamic regression model in the case of non-stationary sequences to explore the causes and influencing factors of false regression phenomena. For the chi-square goodness-of-fit test of the distribution family, it is concluded that when the sample size is small,the error of gradual distribution is large, and Monte Carlo test is recommended.

关 键 词：蒙特卡洛 统计量 假设检验 虚假回归 

分 类 号：O159[理学—数学]