失拟分析在曲线拟合中的应用

The Application of Lack of fit Analysis in the Curve Fitting.

目的 回归分析中 ,曲线拟合的好坏 ,取决于我们选择的期望函数模型是否适宜。对于具有重复试验数据的情况 ,可通过失拟分析来对模型作出评价。方法 用失拟分析方法对含有重复试验数据的期望函数进行失拟检验。结果 若未发现失拟 ,则关于方差的失拟分析已达到它的目的 ;若存在失拟 ,则应试图找出原由 ,并相应地修改期望函数。结论 失拟分析方法值得在对于具有重复试验数据的曲线拟合中加以应用及推广。

Objective In the regression analysis,the performance of the fit is up to the choice of the expectation function for the model.In case of the repeated measure data,we can evaluate the model by lack of fit analysis.Methods Using the lack of fit analysis to test the model expectation function under repeated measure design.Results If No lack of fit is found,the choice of the model expectation function is basically adequate.Otherwise,it imply us to find the reason and modify the expectation function.Conclusion We recommend to do the lack of fit analysis during the curve fitting for repeated measure data.