I'm trying to create a preprocessing step where only covariate (continuous) variables are normalized, however Variable supplier passes all variables not just the specified covariates. Is it possible to restrict the content to just a subset of all the variables? Thank you in advance!
so take a look at the linear regression from jmv, you'll see we have separate boxes for covariates and factors. that's the recommended way of handling this. we discourage analyses which infer how variables should be treated based on their types. as per our docs:
Avoid inferring how the user wants to treat the variable based on its type, i.e. avoid an ‘independent variables’ option, where if the user assigns a nominal variable, it is treated as a factor, and if the user assigns a continuous variable, it is treated as a covariate – this is implied behaviour, and users make mistakes.
I think I might not have expressed myself correctly. What I want is another Variable Supplier under a collapse box where only the Covariates appear (they don't necessarily have to be numerical, although they should really). eg. if the user specifies B as a Covariate, then the new box under the collapse box should only have the variable B as an option, right now I have all A, B, and C irrespectively if they are specified as Dependent Variable, Covariates or Factors (again, doesn't matter if they are continous or categorical although it shouldn't). Can this be implemented or should I avoid such an approach?
ah ok ... sorry, i inferred (incorrectly) what i thought you were trying to do.
you're after something like the following from GAMLj3 (i.e. where the user specifies factors and covariates, and only factors appear in the 'Components' supplier further down? [although in your case, you want to do the opposite, with only the covariates appearing])
jonathon
Screenshot 2025-02-27 at 10.17.15.png (74.44 KiB) Viewed 115324 times