The illusion of uninformative priors

Поделиться
HTML-код
  • Опубликовано: 28 ноя 2024

Комментарии • 8

  • @orktv4673
    @orktv4673 2 года назад

    This is a compelling argument, and I agree: there really is no way of choosing the "right" prior distribution based on no data whatsoever, and even if you could that prior would not be consistent.

  • @CBJrocks
    @CBJrocks 5 лет назад +3

    How did we just remove the integrals at 7:40?

    • @gordongoodwin6279
      @gordongoodwin6279 3 года назад

      Because they aren’t necessary, the point is that the area under the curve and the pdf itself are equivalent, he just put them there at first for clarity

  • @rangjungyeshe
    @rangjungyeshe Год назад

    Nice example - not seen it before.

  • @imranh1225
    @imranh1225 4 года назад

    Also at 8:11, why is P(theta) just 1?

    • @OndrejHavlicek
      @OndrejHavlicek 4 года назад +1

      Because it is a uniform distribution and has value 1 everywhere?

    • @gordongoodwin6279
      @gordongoodwin6279 3 года назад

      Yes, p(theta) is proportional to 1 because it’s uniform

  • @hectorlavaux343
    @hectorlavaux343 2 года назад

    The title is quite misleading, since you talk about the principle of indifference. You can define 'uninformative' otherwise, and then the concept becomes meaningful.