Transformations

Box Cox power transformation

The Box-Cox power transformation is a data transformation to achieve a better normality and to stabilize the variance. In MCRA, the transformation parameter p in (yp1)/p is determined by maximizing the log-likelihood function

l(p)=nslog[1nni=1(y(p)i¯y(p))2]+(p1)ni=1logyi

where i indexes the n observations and

¯y(p)=1nni=1y(p)i

is the average of the y(p)i (Box & Cox, 1964) [Box et al., 1964].