error Run Model recipe with dynamic input feature set

Shutao

Hi,

I need help 😢. I am trying to create a template python recipe. The input will be a dynamic dataset, then i will train different models and redeploy the best model at the end. The error occurs each time when I test to introduce a new dataset (the feature name changes). the deployed model tends to get the old feature set and return error saying the feature xx is empty, even though inside the python recipe I always train again (use mltask.guess() to reidentify new feture set schema.

ps. in the end of the python recipe I create a identical dataset (because an output dataset is mandatory for a recipe).

Thanks in advance for your help!