要求回归因子显著时Box-Cox变换的用法

Methods of using Box-Cox transformation to meet the need of the factors being significant in the regression equation

对在需要使用Box-Cox变换,且要求引入的回归因子均显著的情况下,如何使用Box-Cox变换的问题给出了3种方法,并对3种方法及其计算结果的一些性质进行了理论上的分析与比较。结果表明,3种方法在计算量和计算精度上互有差异,所估计的参数又各有其性质,可根据实际情况选择使用。最后通过对泡桐片崩解试验数据的处理结果验证了3种方法的有效性。

In order to deal with some data,we need to use Box-Cox transformation and to meet the need of the factors being significant in the regression equation.In this situation,the question is how to use Box-Cox transformation.This paper presents three kinds of the methods that they can be used to solve the question,and studies some characters of the methods.Because of some differences in efficiency of computation,error of regression and characters estimate parameters,one of the methods should be selectted according to the factual situation and the need.The validity of the methods is demonstrated by the result to apply them in dealing with the data of Paulownia pill disintegration test.