zhhcxp237 发表于 2010-5-20 23:56 
在很多书上引入随机变量的数学期望时,分布函数虽然能全面地刻画随机变量的统计规律,但有时寻求随机变量的分布函数比较困难,在实际生活中只须知道随机变量的一个或几个分布特征即能结局实际问题,但据我所知,要求数学期望必须要先知道随机变量的分布,试问:如果这个随机变量的分布不知道,是否能求其数学期望,除了一些样本数据求平均值外 [size=156%]
In reality we often use sample mean(nth order moments) to proxy the population mean(nth order moments) if the moments of random variable exist even though we do not know its distribution. And the sample moments converge to its corresponding population moments in large sample. That is the key in estimation.