一种多元均值混合正态分布参数的极大似然估计

Maximum Likelihood Estimation of Multivariate Mean Mixed Normal Distribution Parameters

以Madadi等提出的数学模型与概率密度函数为基础,证明了多元均值混合正态分布的性质,讨论了MMNE分布参数的极大似然估计问题,利用ECM算法对MMNE分布的位置参数、尺度参数、偏度参数进行了参数估计计算,给出了计算方法和迭代公式。实例分析表明,所提出的偏态数据下MMNE分布参数估计模型合理且实用。

Based on the mathematical model and probability density function proposed by Madadi et al.,the properties of the multivariate mean mixed normal distribution are proved,the maximum likelihood estimation of the MMNE distribution parameters is discussed,the position parameters,scale parameters and skewness parameters of the MMNE distribution are estimated by using the ECM algorithm,and the calculation method and iterative formula are given.The case analysis shows that the proposed model for estimating MMNE distribution parameters under skewed data is reasonable and practical.