拟线性回归预测模型的稳定最小二乘解  

Stable Least Squares Solution to Quasi-inearization Regression Model

作　　者：刘学彦[1] 赵建立[1] 相文楠[1] 王慧敏[1] 

机构地区：[1]聊城大学数学科学学院,山东聊城252059

出　　处：《数学的实践与认识》2011年第20期92-97,共6页Mathematics in Practice and Theory

基　　金：国家自然科学基金(10771073)

摘　　要：可线性化回归预测模型通过换元进行线性回归,换元前后的因变量具有异方差性,致使拟线性回归参数的精度较低.运用GL算法给出了此类模型的稳定最小二乘解,提高了参数的估计精度,最后给出了一个应用实例.Linearization nonlinear regression is transformed into model linear regression by substitution, but dependent variables after substitution have heterscedasticity, which makes the precision of the parameters estimated by quasi-linearization regression become lower.In this paper, by using the Global- Local Algorithm,the parameters precision of quasilinearization regression is improved effectively.Finally, an applied example is given.

关 键 词：非线性回归 异方差性 稳定最小二乘解 

分 类 号：O212.1[理学—概率论与数理统计]