Ords, noms – if you want to compute factor levels, the attribute that has ordinal/nominal data type has to be mentioned to avoid continuous value imputations.
idvars – the attributes we need to retain but not used in missing data imputation models are mentioned here.