[Bug]: Feature importances not present in individual ensemble models

[lucazav]

Describe the bug

After I fit an AutoML ensemble for a regression ("ensemble": True in AutoML settings), I get the best model of the ensamble in this way:

best_model = automl.best_model_for_estimator(automl.best_estimator)

If I try to get feature importances for this model in the following way, I get an empty result:

best_model.feature_importances_

Steps to reproduce

No response

Model Used

No response

Expected Behavior

No response

Screenshots and logs

No response

Additional Information

FLAML 2.3.3

[thinkall]

Hi @lucazav , what's the individual model?

[lucazav]

@thinkall best_model.estimator_class returns the following:

sklearn.ensemble._forest.ExtraTreesRegressor

[thinkall]

@thinkall best_model.estimator_class returns the following:

sklearn.ensemble._forest.ExtraTreesRegressor

Hi @lucazav , sorry for the late reply. What's the value of model_history in your code? The default is False. If False, best_model_for_estimator will return a non-trained model.