Open wasciutto opened 1 year ago
Hi Will, I agree that the YAML specification should be complete, including contrasts and other options. My suggestion/hope is that @ryands02 fill out the spec as he writes the save/export function. Happy to be in the conversation about how best to format/structure this.
Adding contrast setup & other remaining options to the L1 YAML file would allow L1 models to be built without using the CLI prompt.
Currently, if user needs to update any of their trial, run, or subject dataframes, they will have to run through the dialogue prompt of build_l1_models again. The user could then potentially select slightly different options that changes their analysis from a previous one they were attempting to duplicate.
I thought about side loading new DFs into an existing, saved gpa object for the above scenario, but it seems like build_l1_models is doing some validation on the DFs that should not be skipped in this way.