Currently, users have to create their own dummy variables for complex factor model constructions. We can allow matrix input, and automated model.matrix call if the input expression returns a formula (like for Spatial covariate objects), and a mapper that handles the mapping to an iid model with fixed precision (like the `factor_models). Need some care to propagate the matrix column names, but can likely be done the same way as the existingfactor_*models. Use example: bru(~ theeffect(~ -1 + a+ b + a:b), ...)`
Currently, users have to create their own dummy variables for complex factor model constructions. We can allow matrix input, and automated model.matrix call if the input expression returns a formula (like for Spatial covariate objects), and a mapper that handles the mapping to an iid model with fixed precision (like the `factor_
models). Need some care to propagate the matrix column names, but can likely be done the same way as the existing
factor_*models. Use example:
bru(~ theeffect(~ -1 + a+ b + a:b), ...)`