奇偶分解及三次抛物线回归的简易算法  

A simple algorithms on cubical parabola regression by the mean of odd-even separation

作　　者：张侠[1] 

机构地区：[1]北京医科大学生物数学与生物统计教研室

出　　处：《数理医药学杂志》1996年第4期299-304,共6页Journal of Mathematical Medicine

摘　　要：对称区间上定义的任一函数均可唯一地表示为一个奇函数与偶函数之和［１］。函数的这一分解实质上是正交分解。巧用这一众所周知的性质于三次抛物线回归，或相似模型上，可使原本三、四个参数的回归模型转化为独立的两个仅含一、两个参数的拟线性回归模型，从而使得参数的最小二乘估计极易在具有统计功能的计算器上实现。且能保持良好的精度。An arbitary function defined on symmetric domain can be always represented uniquely by the sum of an odd funcition and an even function This is practically an orthogonal decompositin When this property is used in cubical parabola regression model or other similar model, the model with four parameters can be transfered to two independent quasilinear regression models, each of them with onlg two parameters. Consequently, it is very easily realized the lest square estimator of the four parameters by calculators with “STAT” function.

关 键 词：奇偶分解 三次抛物线回归 最小二乘估计 拐点 计算器 

分 类 号：R311[医药卫生—基础医学]