It depends on the end goal. If we're predicting apples and oranges, we want to classify both classes correctly. So we ideally should use MCC cuz it is independent of the choice of positive and negative label choice. If we care mostly about one class only, we can go with precision/recall.
Wow, that really helped me in evaluating my master thesis!!!
Very clear explanation. I hope to see more amazing videos in the future. Keep up the good work!
Absolutely, Calin. Thanks!
I feel like I m learning great things, keep it up bro.
I don't usually subscribe to channels after watching one video nor write comments - now I did it because of how good and clear it was
Haha! Thanks a lot!
Great Video Bro... Excellent explanation
Hat's off sir for your clear explanation.
Very clear explanation.
very impressive lecture.. Thank you. 😀
Glad you liked it!
Thank you!!
Good 👍♥️
I have a doubt. How to calculate MCC for multi-class (3 or more)?
very good lecture. i have qus. brother among macro f1-score and MCC which can be better
It depends on the end goal. If we're predicting apples and oranges, we want to classify both classes correctly. So we ideally should use MCC cuz it is independent of the choice of positive and negative label choice.
If we care mostly about one class only, we can go with precision/recall.
👍👍
helpful, thanks!