多元线性模型的广义逆诊断问题  

Diagnostics For Generalized Linear Model Using Generalized Inverse

作　　者：张玲 [1] 柳金甫 [2] 

机构地区：[1]装甲兵工程学院基础部,北京,100072 [2]北京交通大学理学院,北京,100044

出　　处：《装甲兵工程学院学报》2004年第3期5-6,共2页Journal of Academy of Armored Force Engineering

摘　　要：在实际问题中,经常需要考虑多个因变量对多个自变量的相互依赖关系,但在运算过程中经常会碰到奇异矩阵不能求逆的问题.通过推导任意2矩阵的差的广义逆,解决了这一问题,继而得出多元线性模型的参数估计,最后推出数据删除模型异常点的判断依据.we always calculate interdependent connection between cause variables and independent variables in practice. But we often meet strange matrix that can not seek the inverse. In this paper, we discuss the generalized inverse of the difference between the two arbitrary matrixes to solve the problem. Then the estimate of the parameters in the multi linear models is given. Therefore, we get the method to judge the outliers in the data-deletion models.

关 键 词：广义逆 多元线性模型 异常点 数据删除模型 

分 类 号：O212[理学—概率论与数理统计] C8[理学—数学]