试验数据最小二乘回归分析法的改进

Improvement on the method of regression analysis of least square of the testing data

本文提出避开法方程组,改用一系列正交变换,直接求解的方法。此法不仅同样满足最小二乘法原理、极大地减弱“病态“的影响,而且具有规律性强的特点,可以方便地编制计算机程序进行求解。

By the use of a serious of orthogonal transformation this paper puts forward a direct solving method in order to avoid using the normal equation group. This method not only satisfies the principle of least square method and greatly weakens the effect of 'morbid state' but also provides with a characteristic of strong regularity and computer programs can be easily made to find the solution.