Definition: The Box-Cox transformation, below, can be applied to a regressor, a combination of regressors, and/or to the dependent variable in a regression. The objective of doing so is usually to make the residuals of the regression more homoskedastic and closer to normal distribution:
None
None
y(l) = ((y^L) - 1) / l for l not equal to zero y(l)=log(y)L=0 Box and Cox (1964) developed the transformation.
Estimation of any Box-Cox parameters is by maximum likelihood.
Box and Cox (1964) offered an example in which the data had the form of survival times but the underlying biological structure was of hazard rates, and the transformation identified this.(Econterms)
Terms related to Box-Cox Transformations:
About.Com Resources on Box-Cox Transformations:None
Writing a Term Paper? Here are a few starting points for research on Box-Cox Transformations:
Books on Box-Cox Transformations:
- Davidson, Russell, and James G. MacKinnon. 1993. Estimation and Inference in Econometrics. Oxford University Press.
None
