This is a follow-up question of https://stackoverflow.com/questions/41127976/confusing-example-of-nested-cross-validation-in-scikit-learn.
How would one retrieve the best parameters of the inner cross validation loop of the nested CV example linked in that question?. Is it exactly the question raised in this issue: https://github.com/scikit-learn/scikit-learn/issues/6827?
In other words, the best performing model should be chosen from the
GridSearchCV optimized internally within
cross_val_score, is that right?
[Assuming that we had only one experiment instead of the
NUM_TRIALS of the example] Selecting directly
clf.best_params_ (from the outer
clf object) instead of inspecting the fitted
cross_val_score would be wrong, wouldn't it?