When schemas are generated based on variable data, for example as outputs from pivot recipes, they can sometimes end up inconsistent between training and prediction datasets for models. It would be really helpful to be able to copy schemas between datasets or link them dynamically, ensuring prediction and training sets have consistent schemas. Then, any missing values could be automatically left null or assigned static values, allowing the flow to build even as schemas are updated with changing values.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.