非线性回归方程选择的一种新方法

A new method to select equation of nonlinear regression

指出非线性回归方程选择中存在的问题 ,即先给定回归方程的表达式而后确定方程中的待定参数 ,这种回归方程的选择是盲目的 .提出用合类偏相关系数的大小作为选择方程的依据 ,并给出计算该系数的优化算法 .经实验数据考核对比 ,表明该方法能根据实验数据本身的大小和特征 ,有根据地得到满意的回归方程 。

It is pointed out that selecting the equation of nonlinear regression is blind because the expression of regression equation is determined before its parameter is determined. Based on the whole class partial correlation coefficient the equation of nonlinear regression is selected, and the optimize algorithm to calculate the coefficient is given. It is suggested that the appropriate expression of the regression equation could be got according to the quantity and characteristic of the data, so that the blindness in selecting regression equation is coped with.