The slide at 4:00 could be wrong. It should be "Traning set, Training set, Test set" for the first iteration and "Traning set, Test set, Traning set" for the second iteration. Also, the slide at 7:22 could be incorrect for the nested cv.
Thank you for making a video on this! Very valuable! It took me some time to find out there exists a lot of research on proper simultaneous error estimation (model selection) and hyperparameter tuning!
In the slide at 6:37, the best hyperparameter ("n_estimators=5") was chosen for the third iteration (third fold). But what happens if the hyperparameters on the first and second folds are different? I understand that we can evaluate the overall performance using different hyperparameters in the other folds of the outer loop. But, at the end of the day, which hyperparameter we choose to produce a final deployment model?
@@ploomber I think the format is well done. Well organized slides and reproducible code snippets. Will get back when I have additional remarks or concerns. Cheers!
![ILLSLICK - KILLSHOT REMIX [FULL VERSION]](http://i.ytimg.com/vi/RIK8RrqIAzE/mqdefault.jpg)
This video + the ploomber article are the best resources I found on the topic. They are both complementary ! Thank you.
The slide at 4:00 could be wrong. It should be "Traning set, Training set, Test set" for the first iteration and "Traning set, Test set, Traning set" for the second iteration. Also, the slide at 7:22 could be incorrect for the nested cv.
yeah, confused the f out of me
Thank you for making a video on this! Very valuable! It took me some time to find out there exists a lot of research on proper simultaneous error estimation (model selection) and hyperparameter tuning!
Thank you for this video, i learned so much, you are amazing teacher
In the slide at 6:37, the best hyperparameter ("n_estimators=5") was chosen for the third iteration (third fold). But what happens if the hyperparameters on the first and second folds are different?
I understand that we can evaluate the overall performance using different hyperparameters in the other folds of the outer loop.
But, at the end of the day, which hyperparameter we choose to produce a final deployment model?
Cool insights here!
What do you think of the video format? Any feedback for us?
@@ploomber I think the format is well done. Well organized slides and reproducible code snippets. Will get back when I have additional remarks or concerns. Cheers!
@@wyclifeoluoch9376 Awesome! Please join our community to keep in touch! ploomber.io/community