The original paper (Guerrero (1993)) which discusses Guerrero's method to find the Box-Cox lambda is given only for strictly positive data.
So I believe we should issue an error message (or warning) to ensure that negative data or a mixture of positive and negative data are not allowed (or use with caution). Or else we might have to modify the guer.cv() function to handle these types of data.
Also, modifying the function to better handle X<=0 would justify a new function / method. Using guerrero() (and consequently, guer.cv()) should give you the Guerrero (1993) method.
The original paper (Guerrero (1993)) which discusses Guerrero's method to find the Box-Cox lambda is given only for strictly positive data.
So I believe we should issue an error message (or warning) to ensure that negative data or a mixture of positive and negative data are not allowed (or use with caution). Or else we might have to modify the
guer.cv()
function to handle these types of data.