Stop Using Accuracy! 4 Essential Metrics for Imbalanced Data

Поділитися
Вставка
  • Опубліковано 23 січ 2025

КОМЕНТАРІ • 13

  • @hikeaway1596
    @hikeaway1596 12 годин тому +4

    🙏👍 I always learn new useful things from you videos, which I didn't know exist. Thanks for doing this and keep up a great work!

  • @RohitSharma-uw2eh
    @RohitSharma-uw2eh Годину тому

    Why hindi audio track is not available

  • @hashimalikhan775
    @hashimalikhan775 5 годин тому

    Great again as always. Kindly do a video on Meta regression.

    • @yuzaR-Data-Science
      @yuzaR-Data-Science  4 години тому

      Thanks mate! Meta-regression is an interesting topic. But I don't use in everyday job. That means, this video will not be coming soon. I already have a long list with I do often and folks request often. So, I am sorry to disappoint, but I'd rather be honest with you, then let you wait and hope and don't deliver. Hope for your understanding!

  • @yuliaegorova589
    @yuliaegorova589 3 години тому

    I wonder if you can do video on different threshold metrics: I usually never pick an arbitrary threshold like 0.5 to convert to binary : but I compute the reshoots based on ROC curve (Yonden index )or TSS metrics . I wonder if you ever do smth similar in your job :)

    • @yuzaR-Data-Science
      @yuzaR-Data-Science  3 години тому +1

      Hi Yulia :) Did you miss my video on ROC Curve? I just have done one in December. And it's about exactly what you ask for: using youden index and other metrics like missclassification rate to find optimal thresholds. Or did you see it and you mean something different?

    • @yuliaegorova589
      @yuliaegorova589 3 години тому

      @@yuzaR-Data-Science oh let me rewatchit! it looks like i get my answer from that video:)

    • @yuzaR-Data-Science
      @yuzaR-Data-Science  2 години тому

      I hope 🤞 you will