Maximum Likelihood Examples

[@danielgraaf5042]

Best tutorial on KZbin! Thank you! Fast and to the point! Kudos!

[@brookehope3709]

The most helpful MLE explanation. I suck at math and every other video bewildered me but this makes MLE a lot more clear. Thank you!

[@YYZ722]

Have watched several videos about MLE. This is the one which I believe makes the most sense.

[@UpSwinging]

Very fast-paced but still straightforward and clear-cut content.Congratulations on the delivery.

[@AakarshNair]

Really appreciate the use of simple and concrete examples in the beginning

[@toncao3723]

Better than lecture from my uni

[@muhammedcinsdikici]

Thank you for giving us the concept of MLE in a simplest form. The tutorials of the MLE are more complex to follow.

[@PieterAbbeel]

Minor typo notice: for the binomial distribution, it should be (1 - \mu)^4 \mu^2 (rather than (1 - \mu)^3 \mu^2). This results in \mu = 2/6 (rather than 2/5).

[@Uma.gupta123]

Before watching this video MLE was tough for me. Thanks for making this video 👍

[@Blockman367]

Thank you for the video, I have looked on the Internet but not found a concrete and easy example until i fond your video

[@EfeCevher]

The simple explanation is always the best one, thnx Pieter.

[@ruanrichter]

I think on the third line you write down of the first example it should be (1-mu)^4 instead of (1-mu)^3. But thank you very much! Great video :)

[@PieterAbbeel]

Thank you!!

[@dianasylvia2705]

yes, i found that too

[@gregstauffer8922]

Very informative. Clearly showed the concept. Thank you for publishing!

[@pivoodon]

best video of the day. Thanks a lot! All videos I've seen go through the equations and I just want to see an example of how the equations are applied.

[@MrAntiKnowledge]

So easy to understand. It was so cryptic when my prof explained it.

[@Sapphireia]

Awesome, love that it's fast paced!

[@udupi123456]

Best tutorial explained very well...

[@toebeeee]

great video and introduction to ML. minor complaint: the ML estimator for the binomial distribution is wrong since the probability of observing a 1 should be 2/6...the likelihood function should say: (1-u)^4*u^2....

[@mohsenhs]

Thank you Abbeel , Much appreciated. very helpful and excellent explanation.

[@esheskinner9508]

Thanks alot for such a simple ,precise and well explained tutorial

[@jorgec7028]

excellent explanation, so easy to understand it with your video ¡muchas gracias!

[@ashumohanty566]

it is the best vdo thank you sir

[@harismalik8806]

OMG...! Sir you make it so easy thanku soo much!

[@aminghanooni4240]

Thanks for replying my message. I am wondering whether you can present videos for Method of Moments, Least Squares principle and Bayesian Estimation. It is defined so complex in the literature. I really liked how you explained in a simple way through the examples. :-)

[@aminghanooni4240]

Really Perfect. God bless u. Thanks for everything

[@josemonge9048]

nice video! even you have a typo as Ruan Richter points out, it is a well established derivation that for the binomial distr. the MLE is just the sample mean -> 2/6, in this case.

[@davidfield5295]

Very good video

[@angelaskari1124]

Thanks a lot Sir. Very well explained.

[@AGhoreyshi]

Thank you for your "thank you", Evan (I am speaking on behalf of Pieter). Also, thanks Pieter. Good video. My only suggestion would be to use pretty colors like those used in Khan Academy.

[@acelyacan4697]

Well explained. Thanks :)

[@kunjaai]

Thanks a lot 4 your nice lecture ......

[@kimbapkidding6592]

Thank youu, do you have video example likelihood for normal distribution???

[@amanirouihem58]

thanks a lot , short and perfect

[@KG-iy5ll]

Didn't lost 1 term of 1-μ in the first example ?

[@prof_shixo]

Which course was this video from? Is the playlist available?

[@HariAnantharaman]

Thanks - Excellent Focussed Tutorial

[@MrHugosky1]

Great Lesson! Thank you

[@daveh3930]

great content, thank you!

[@meydiarachma4458]

Thank you sir. You rock

[@riccardomoro8159]

But if I find that the second derivate doesn't give me a maximum what do i have to do? in this case the maximum doesn't exist right?

[@TarunKumar-en8si]

While calculating likelihood you are multiplying the individual probabilities. I think we can only do that in case where individual trials our independent events according to our model. Or am I missing something?

[@bend.4506]

just what I was looking for

[@antoninayun1538]

Thanks a lot! really good explanation:)

[@gustavogrossi8019]

hey! great video thanks! in the second exercise, shouldn't d(43log(lambda))/d(lambda) be equal to 43/lambda*ln10, instead of 43/lambda as you pointed?

[@juancuadra3697]

great tutorial. Thank you

[@JustCheckingMusic]

Thanks, helped a lot!

[@user-mb3mf2og9k]

6 samples, 2 ones, so p(x=1) is 2/6. no derivative needed. right? and for Poisson 5+9+3+12+14/5=43/5

[@tshiamomogaadile9705]

you've just saved my life

[@thomaryanfajri8782]

great tutorial vid

[@dr.vinodkumarchauhan3454]

perfect, thanks

[@harshitajain1628]

sir in general case what's the mle of binomial distribution?

[@junyizhong]

crystal clear!

[@abhinavkilaru8294]

very helpful

[@panklbj]

I understand how maximum likelihood estimator is derived by watching your video but I just don't understand why in this way we can estimate the estimator. Why multiplying all the pdfs and maximize it can give an estimate of a certain parameter?

[@PieterAbbeel]

Hi Yuyui, thanks for your note! Agreed the video only explains the ML procedure, and not the rationale as to why the ML procedure might be a desirable way to estimate parameters. The rationale behind ML is to find the parameter setting that maximizes the probability of the data. The probability of the data equals the product of all the pdfs (assuming each data point was drawn independently). With lots of data and a good parameterization of your pdf, this can work quite well. With very little data, it might overfit, and you might consider using a prior. Cross-validation can help you determine whether you might be overfitting or not.

[@dinafarfosheh2107]

thanks alot.. very helpful ^^

[@realcirno1750]

thank you

[@annalam8624]

great video! thank you!! :)

[@AdarshMammen3]

Thank you!

[@PETAJOULE543]

Quite robotic mic :D anyways this was helpful and simplified my understanding of ML estimation

[@PieterAbbeel]

Thanks!

[@photinoman]

Nice!

[@erolxtreme5081]

mu is the mean?

[@benchipperfield7989]

holy shit i get it now :o thanks pieter

[@annarauscher8536]

THANKS

[@mr.z9161]

The first equation you write was run, the power of (1-u) should be 4. Be careful, buddy.

[@rwebo4955]

i notice that too when i tried to do it by myself here.. and its untill second derivative that you find MLE, not the first one

[@sujatabhaumik1506]

Alan Lwanga x9xx8xxx9CSX's CDs x97FCC 87xx

[@sujatabhaumik1506]

Alan Lwanga xu cultigens diphthongs you're

[@sujatabhaumik1506]

ziqi zhu "y CFCs y 7 Chateaubriand FRC utterances AF ycx" x y y

[@sujatabhaumik1506]

xy d8

[@ojaspandey2473]

Should be (1-u)^4

[@GeorgePapageorgakis]

I think you have used too much microphone boost and the sound is distorted-terrible :o Nice examples though.

[@raoufzanati7532]

king

[@jauekya]

Easy peasy japanesy

[@cameronlarson4046]

You go too fast and your video is annoyingly choppy.

[@PieterAbbeel]

Thanks!