Box-Cox: calculate lambda for negative data


My question is how to calculate lambda for the original series y(t), if y(t) is negative or has zero's?
Because from forecast (fpp2) forecast source: R/guerrero.R says
guerrero: "Guerrero's method for selecting a Box-Cox parameter (lambda) is given for strictly positive data" and
bcloglik: Modified version of boxcox from MASS package "x must be positive".
And in feasts (fpp3) guerrero is similar feasts source: R/guerrero.R

Thus, I cannot calculate lambda for negative y(t), and use eqn(3.1), eqn(5.2) and eqn(5.3) in fpp3 (which allow for negative values of y(t) provided lambda>0), if y(t) the original series is <=0.


This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.