Why is a likelihood not a probability distribution?

Поділитися
Вставка
  • Опубліковано 28 лис 2024

КОМЕНТАРІ • 30

  • @NuclearSpinach
    @NuclearSpinach 2 роки тому +7

    I'm multiple years into studying graduate statistics and I've never heard this described so simply. People jump through so many hoops trying to describe what likelihood is, but you perfectly boil it down to the essence here. I've never heard it described so simply. Thank you.

  • @philmccavity
    @philmccavity Рік тому +1

    Like others, after years of studying statistics, and reading your book on Bayesian statistics and enjoying every page, I still find myself watching these videos. The pace is perfect, the content, the explanation. I remain extremely grateful for this free resource.

  • @ehdghgka
    @ehdghgka 4 роки тому +9

    I've studied statistics for like 2 years. Finally I got the difference between likelihood and probability distribution.

  • @hassamsheikh
    @hassamsheikh 5 років тому +10

    from, 3 days, I was trying to wrap my head around this concept. Finally an eight minute video did the magic

  • @BobBruer
    @BobBruer 4 роки тому +10

    After years of study at superb institutions... after publishing complex papers using advanced multi-variate modelling techniques... I finally understand 'likelihood'. Thanks so much! One criticism... in the first graph, it would have come to me with fewer viewings if the graph looked like a true density (area under the two heights) instead of two straight vertical lines.

    • @mech_builder7998
      @mech_builder7998 4 роки тому

      The final point you make is very helpful - "in the first graph, it would have come to me with fewer viewings if the graph looked like a true density (area under the two heights) instead of two straight vertical lines."

  • @こよい-e7n
    @こよい-e7n 6 місяців тому

    Great! You let me understand the concept confusing me for a long time! Thank you!

  • @suyashshandilya9891
    @suyashshandilya9891 5 років тому +20

    This video is misplaced in the playlist. It should be placed before the video on equivalence relation.

  • @aroonalok
    @aroonalok 3 роки тому

    This is the best video on this topic so far.

  • @vaibhav_uk
    @vaibhav_uk 2 роки тому

    Grateful that this video exists

  • @nikolaspediotidis2804
    @nikolaspediotidis2804 5 років тому +11

    2:09 "alternatively if k is equal to 0* (not 1)..."

  • @davidpaganin3361
    @davidpaganin3361 6 років тому +5

    Superb -- extremely helpful -- much appreciated!

  • @waliedahmed5030
    @waliedahmed5030 4 роки тому +1

    Brilliant and simple. Thank you, sir.

  • @bunkerputt
    @bunkerputt 6 років тому +1

    First view! Thanks Ben for the amazingly clear videos. Awesome work! 🏆

  • @vishnuvardhan-md1ux
    @vishnuvardhan-md1ux 3 роки тому

    At 4:39 the graph you plotted for theta vs p(x=1/theta) should be normal right? assuming p(x=k)=theta^k*(1-theta)^k, then for p(x=1)=theta*(1-theta) then for all values of theta with x=1 lies between [0,1] with a mean of 0.25 and symmetric curve.

    • @vishnuvardhan-md1ux
      @vishnuvardhan-md1ux 3 роки тому

      sorry i got you the p(x=1) becomes just theta after substitution.

  • @JNCressey
    @JNCressey 5 років тому +1

    Why do the two arguments swap when Pr is swapped for L?

  • @samirajodeir1285
    @samirajodeir1285 3 роки тому

    Very nice explanation!

  • @zeeshan008x52
    @zeeshan008x52 5 років тому

    Very well explained sir. Hats off

  • @wavejoejoe
    @wavejoejoe 4 роки тому

    well explained, much appreciated video!

  • @homataha5626
    @homataha5626 4 роки тому

    It was perfect, thank you

  • @TylerMatthewHarris
    @TylerMatthewHarris 4 роки тому

    Thank you!

  • @hey_im_random
    @hey_im_random 4 роки тому

    THANKS!

  • @joshcoulter9017
    @joshcoulter9017 3 роки тому

    I believe Episode 16 should be, chronologically, in before Episode 14.

  • @teelee3543
    @teelee3543 6 років тому

    why the origin of the x axis does not start from zero ?

    • @gauthambekal235
      @gauthambekal235 4 роки тому +2

      It doesn't matter its just a notation or label.Instead of writing 0 you could have just labeled x with T and H representing tails and heads.

  • @AnnevanRossum
    @AnnevanRossum 4 роки тому +1

    L(t|x=0)+L(t|x=1)=1 though, t=theta. Easily visualized by drawing the graph for both cases. The triangles form a unit square.

    • @AtmoStk
      @AtmoStk 2 роки тому

      You need to study Bayesian basics.

    • @AnnevanRossum
      @AnnevanRossum 2 роки тому

      @@AtmoStk I did. Prove me wrong please.

    • @Justin-zw1hx
      @Justin-zw1hx Рік тому

      This is what I was thinking too, he did both x =0 and x=1 for probability, but he only did x =1 for likelyhood, which does not make sense.