Box and Cox (J Royal Stat Soc Ser B 26:211–252; 1964) suggested a family of power transformations to improve additivity, normality, and homoscedasticity of observations. According to the review by Sakia (Statistician 41:169–178; 1992), the Box-Cox transformation and its improved versions are described and then their applications to linear or nonlinear regression are briefly mentioned as follows.