Box-Cox Transformation with Categorical independent variable in model

[bhr-q]

Hello all,

Is it correct to use the code below to obtain the λ in a Box-Cox transformation?

I’m unsure because my model includes an independent categorical variable (pre_score) plus a few categorical and continuous confounders (age neuro_scale sex).

I was looking for a way to include a class statement to account for the categorical nature of these variables, but it seems that PROC TRANSREG does not support the class statement.

Is that necessary to include the categorical nature of these variable?

%let c = 1.45;
proc transreg data=tmp1      ss2 details plots=(boxcox);
   model BoxCox(fim / parameter=&c  
                         convenient lambda=-2 to 2 by 0.05) = identity(pre_score age neuro_scale sex);
run;

Thanks,

[Ksharp · Accepted solution]

If you really want include some category variables in PROC TRANSREG ,
You could try to transform them into 0-1 variables to form a Design Matrix(include continuous variables),
and feed this Design Matrix into PROC TRANSREG .
Check @Rick_SAS blogs:
https://blogs.sas.com/content/iml/2016/02/24/create-a-design-matrix-in-sas.html