Prediction problem for component models while using ensembling

[D3TR0]

Hi,

I have a problem with using predict functions for component models. When i use the main model everything is ok

model_reg = AutoML()
model_reg.fit(X_reg, y_reg, task="regression", ensemble=True, time_budget=120)
model_reg.predict(X_reg)

However, if I try to do this for component models I get all sorts of errors, depending on the model
ensemble_models = model_reg.model.estimators_
ensemble_models[0].predict(X_reg)

ValueError: The feature names should match those that were passed during fit.
Feature names must be in the same order as they were in fit.

ensemble_models[2].predict(X_reg)
ValueError: train and valid dataset categorical_feature do not match.

ensemble_models[3].predict(X_reg)
ValueError: DataFrame.dtypes for data must be int, float, bool or category. When categorical type is supplied, The experimental DMatrix parameterenable_categorical must be set to True. Invalid columns:Gender: object, Education Level: object

[D3TR0]

I think I have found a solution to this problem. Before the prediction for the component model, you need to do transformations using the final model.

X_transformed = model_reg._state.task.preprocess(X_reg, model_reg._transformer)
ensemble_models[1].predict(X_transformed )

However, it seems to me that this behavior is not intended