基于StN分布下联合位置与尺度模型的极大似然估计  

Maximum Likelihood Estimation for Joint Location and Scale Models of the Skew-t-normal Distribution

作　　者：吴刘仓[1] 马婷[1] 戴琳[1] 

机构地区：[1]昆明理工大学理学院,云南昆明650093

出　　处：《应用数学》2013年第3期671-676,共6页Mathematica Applicata

基　　金：国家自然科学基金资助项目(11261025;11126309);云南省自然科学基金资助项目(2009ZC039M;2011FB016;2011FZ044)

摘　　要：在社会和经济等领域中存在大量的异方差数据,而且人们非常关注方差的变化,所以在研究社会经济现象时方差建模与均值建模同等重要;相对于对称分布,偏态分布更能获得更全面准确、更及时有效的信息,文章基于以上两点,研究提出基于偏t正态分布(StN)的联合位置与尺度模型,并给出该模型参数的极大似然估计,模拟和实例研究结果表明该模型和方法是有用和有效的.In social and economic fields,there are a lot of heteroscedastic data,and we are very concerned about the change of the variance, thus modeling of the variance can be as im- portant as that of the mean in the social economic phenomenon. On the other hand, compared with the symmetrical distribution, skewness distribution can obtain more accurate and more timely information. Based on the above two points, we propose a joint location and scale mod- els based on the skew- t -normal distribution,and investigate the maximum likelihood estima- tor of this model. Simulation studies and a real example show that this model and method are useful and effective.

关 键 词：StN分布 联合位置与尺度模型 极大似然估计 

分 类 号：O212.1[理学—概率论与数理统计]