Lp空间上线性回归方程回归系数的估计

本文讨论了多元线性回归方程参数的L^p估计问题。首先采取消除方程中参数“。的技术，化参数估计为对回归系数的估计，然后进一步将问题转化为一个单目标数学规划模型，利用能自动搜索最优解的电脑软件，十分方便地求出规划模型的最优解。特别对一元回归方程，在消除参数a0后，还可以直接利用普通一元函数求极值的方法，求出回归系数的估计。

Estimation on Regression Coefficient of Linear Regression Equation in Lp Space

This paper discusses the problem on parametric estimation of multivariate linear regression equation in L^p space. First, we use the technology dispelling parameter a0 in equation, and simplify estimation of parameter into estimation of regression coefficient. Then, we transform this problem into a mathematic programming model of single goal. Further using computer software of searching optimum solution of programming model voluntarily, we can obtain optimum solution of programming model very conveniently. In particular, to a variable regression e quation, after dispelling parameter a0, we can get estimation of regression coefficient by ordinary method of extracting extreme value on a variable function direct ly.