#77: Scikit-learn 74:Supervised Learning 52: Project: Predict Greenhouse gases conc.

Поділитися
Вставка
  • Опубліковано 31 жов 2024

КОМЕНТАРІ • 8

  • @ganesanelumalai3596
    @ganesanelumalai3596 2 роки тому

    Thank you for the detailed explanation with real-world data.

  • @Mohammad-vr9dj
    @Mohammad-vr9dj 2 роки тому

    Hello sir, Thanks for your useful video.

  • @luisbarba9532
    @luisbarba9532 2 роки тому

    One question, how would you implement a heteroscedastic gaussian to help determine noise variance?

  • @a-nahamizunbinmaamor1834
    @a-nahamizunbinmaamor1834 2 роки тому +1

    Hi Sir, you have done such great explanation in this video. Your series video on Gaussian Process Regression very helpful.
    However, I have a curiosity on how to decide to use the best kernel for the any data?
    and
    how can I tune the value from kernel parameter like k1, k2, k3 and k4 to get the best fit for any dataset?

  • @Mohammad-vr9dj
    @Mohammad-vr9dj 2 роки тому

    Sorry, I have a question. If our dataset will be multidimensional, how can we deal with this issue? you can imagine we have the dataset with dimensions x_train:(200,1280), y_train:(200,1280), x_test:(24,1280) and y_est:(24,1280).

    • @Mohammad-vr9dj
      @Mohammad-vr9dj 2 роки тому +1

      @@learndataa thanks for your response 👍
      Actually, 1280 can't be reduced. I should define a loss function and compare the real y and the prediction.
      I faced some problems when I wanted to import my dataset. My actual dataset has below dimension:
      X_train : 240, 64, 10, 2
      Y_ train: 240, 1280
      X_test: 24, 64, 10, 2
      Y_test: 24, 1280
      I changed my dimensions at first to the ones which I mentioned in the previous comment. But, I shouldn't reduce them.
      If it is possible guide me, please 🙏
      And let me ask you about convolutional kernel; why didn't you define convolutional kernel?

    • @Hamza-vk6sc
      @Hamza-vk6sc 9 місяців тому

      The same question bro,
      did you find solution to your problem ?