The most intuitive description of conjugate priors and credible intervals going beyond bayes theorem......a great lecture and a brilliant teacher indeed.
One thing that I used to get confused about was how the prior distribution for any parameter is chosen when doing bayesian regression. This video really answered the question at this time. I am grateful TH-cam has a lot to learn if one is willing to do so because of so many teachers putting up videos. Thank you for the efforts creating the playlists.
Thank you for this wonderful video. Thanks to your video, I finally understood what I unfortunately never understood in months of lectures from my professor.
Thanks for the video! I'm 18 and really interested in statistics for scientific research and this has been a great introduction to bayesian statistics! Really nice graphics to follow along :)
Hi, thanks for the video and do appreciate your effort in doing all the other videos. They have been very helpful. I know you are currently working on the survival analysis video series, and I want to plea if you can do a video on the application of this Bayesian method, using a real world data (maybe health data) and especially in conjunction with the MCMC method. My interest is to understand its application to a data with various variables.
We all did. We just didn’t notice it. We were pretty sure we liked hamburgers and so we did the experiment at the local hamburger joint which wasn’t that good. So our posterior opinion is that maybe hamburgers are not that great. Now we have a new prior and we can do another experiment at a different hamburger place. That will give us a new expectation. Now, if we had access to quantitative of our thought processes, we could use the formula. But we don’t need the formula and we don’t need the name. Bayes was describing what we do all the time.
Thank you very much for well articulated video on introduction of Bayesian Data analysis. Your efforts can not go unappreciated. If possible, could you share more videos on the application of these in health inclusive of application of MCMC
Bayes theorem in grade 9? I didn't even know what it was during my bachelor and masters i only hear my teaching mentioned it. I understood it better when I took a course called Bayesian data analysis last fall at bowling Green state university then I understood what it really was
The 'interval' refers to the confidence interval, expressed as some multiple of the standard deviation of the distribution. The most commonly reported confidence intervals are 90%, 95% and 99%, which correspond to intervals of + or - 1.64, 1.96 and 2.58 standard deviations, respectively. To say that "95% of similarly sized intervals from repeated samples of size n will contain theta" is to say that if you drew 100 samples (of size n) from the same population and calculated each sample's mean, then 90 of those sample means would be within 1.64 standard deviations of the population mean, 95 would be within 1.96 standard deviations of the population mean, and 99% would be within 2.58 standard deviations of the population mean.
At 17:00 How do you know in general the likelihood equation in a real life problem? Usually you don't the theoretical process producing the data, you just have an unknown sample. Here you are assuming both the a priori equation and also the likelihood equation. I can only understand an assumption for the likelihood if we have a bernouilli process or for the sample mean (because of the central limit theorem).
For example in clinical research you define your apriori distribution from related pharmaceutica and their testing. Does this help answering your question or did I misunderstood you?
Where is data in P( theta | data )? I mean, how is the single coin toss result taken into account in the posterior distribution in your example and how do you infer that the P( data ) is 0.5?
Hi I recently ran several Bayesian independent samples t tests using the informed prior vs the default prior. I understand why the same analyses using the informed prior gives a bigger Bayes factor compared to the default, but I didn’t expect the effect sizes to all be smaller (and credible intervals narrower) when using the informed prior. Does anybody know why this might be?
FREQUENTIST Confidence Interval Repeated sampling. Each contain N data. 95% of all samples contain value X within a specific interval. BAYESIAN Credible Interval 95% chance that X value is within a specific interval.
Bayesian*: granted "this" prior distribution and "this" likelihood function. So it's not that beautiful of an answer either. Though frequentist CIs also require quite some assumptions
Por alguna extraña razón las mujeres le dan más valor a lo que escuchan que a lo que esta escrito. Y valgan verdades, la sociedad es femenina por definición Por alguna extraña razón la gente le da más valor a las promesas electorales de aquel que se autoproclama defensor de tus derechos y te pide el voto a cambio de mejorarte la vida. Sin embargo, luego de unos años, las únicas vidas que han mejorado son las vidas de aquellos que se habían autoproclamado. Si la política consiste en engañar a la gente y a cuanta más personas engañes mejor. Pregunto, ¿tu te sientes parte del problema o de la solución? - IMV P2rtido Polític0 en WordPr3ss -- El primer partido político con funcionamiento interno verdaderamente democrático en la historia de España. -- Sin machitos alfa, sin capataces a lomo de caballo blanco, sin tonto pollas, sin chulo playas, sin cuadros, sin mi3rdas
The most intuitive description of conjugate priors and credible intervals going beyond bayes theorem......a great lecture and a brilliant teacher indeed.
You kidding...
I rarely ever comment on anything but oh my god. Thank you SO MUCH for this.
one of the best videos out there. My PhD dreams thank you so much...
One thing that I used to get confused about was how the prior distribution for any parameter is chosen when doing bayesian regression. This video really answered the question at this time. I am grateful TH-cam has a lot to learn if one is willing to do so because of so many teachers putting up videos. Thank you for the efforts creating the playlists.
Guys, you are perfect to explain all things relating to statistics
Thanks!
it is explained much clearer than the other similar videos
Congratulations on creating an exceptionally clear explanation of the basics of Bayesian statistics!
You should be proud of every student you help with these videos!
Thank you for this wonderful video. Thanks to your video, I finally understood what I unfortunately never understood in months of lectures from my professor.
You might just have saved my PhD and I love you
Thanks for the video! I'm 18 and really interested in statistics for scientific research and this has been a great introduction to bayesian statistics! Really nice graphics to follow along :)
Hi, thanks for the video and do appreciate your effort in doing all the other videos. They have been very helpful. I know you are currently working on the survival analysis video series, and I want to plea if you can do a video on the application of this Bayesian method, using a real world data (maybe health data) and especially in conjunction with the MCMC method. My interest is to understand its application to a data with various variables.
Thank you! But honestly, who did Bayes Theorem in grade 8 or 9? I didn't even do it in depth in my bachelors..
We all did. We just didn’t notice it. We were pretty sure we liked hamburgers and so we did the experiment at the local hamburger joint which wasn’t that good. So our posterior opinion is that maybe hamburgers are not that great. Now we have a new prior and we can do another experiment at a different hamburger place. That will give us a new expectation. Now, if we had access to quantitative of our thought processes, we could use the formula. But we don’t need the formula and we don’t need the name. Bayes was describing what we do all the time.
Thank you very much for well articulated video on introduction of Bayesian Data analysis. Your efforts can not go unappreciated. If possible, could you share more videos on the application of these in health inclusive of application of MCMC
great video. would be interesting to see how you extend this lesson to situations in which you don't have a conjugate prior
A Great lecture. Thanks a lot!
Thank you, very clear explanation and very nice visuals!
I want to learn sufficent statistics from you. Please make a video on sufficient statistics
My professor could never... thank you!
Love it and also love your correction to Dirichlet!
More on bayesian inference in prospect of research
Bayes theorem in grade 9? I didn't even know what it was during my bachelor and masters i only hear my teaching mentioned it. I understood it better when I took a course called Bayesian data analysis last fall at bowling Green state university then I understood what it really was
Really good explanation. Thanks for this!
3:09 What does "similar sized intervals" means please? As in "95% of similar sized intervals from repeated samples of size n will contain θ".
The 'interval' refers to the confidence interval, expressed as some multiple of the standard deviation of the distribution. The most commonly reported confidence intervals are 90%, 95% and 99%, which correspond to intervals of + or - 1.64, 1.96 and 2.58 standard deviations, respectively. To say that "95% of similarly sized intervals from repeated samples of size n will contain theta" is to say that if you drew 100 samples (of size n) from the same population and calculated each sample's mean, then 90 of those sample means would be within 1.64 standard deviations of the population mean, 95 would be within 1.96 standard deviations of the population mean, and 99% would be within 2.58 standard deviations of the population mean.
@@EagleSlightlyBetter Thanks very much. What does "similarly sized" mean here please? "Similar" to what "size"?
Hi
Have you done anything about latent class analysis and cluster analysis
Many thanks your intuitive teaching is amazing!
At 17:00 How do you know in general the likelihood equation in a real life problem? Usually you don't the theoretical process producing the data, you just have an unknown sample.
Here you are assuming both the a priori equation and also the likelihood equation.
I can only understand an assumption for the likelihood if we have a bernouilli process or for the sample mean (because of the central limit theorem).
For example in clinical research you define your apriori distribution from related pharmaceutica and their testing.
Does this help answering your question or did I misunderstood you?
Hey! Great video, wish you'd make another one about MAP and Bayes factors (I suppose you could fit them into 1 video).
I love this video. Thank you 😊
Great video. I found it very helpful. Thanks.
Absolutely brilliant!
Where is data in P( theta | data )? I mean, how is the single coin toss result taken into account in the posterior distribution in your example and how do you infer that the P( data ) is 0.5?
Amazing! Thanks a lot! I wish there were more videos on this topic 🤓
Very good video. How was the value of the normalization constant (0.5) determined?
So helpful, thank you!
a great lecture.
You are the best!
Great! Thank you!
9:16 Why is P(data) a constant?
this is the best explanation of the subject i have found on youtube 😂😂
Can we say that Marginal probability and Prior probability are the same ?
Hello sir. Please what is the best statistical model on e-view to be used when the number of observations are small.
Thank you! (I'd heard of Dirichlet in the context of boundary conditions for the Navier Stokes equations when modelling fluids.)
Hi
I recently ran several Bayesian independent samples t tests using the informed prior vs the default prior. I understand why the same analyses using the informed prior gives a bigger Bayes factor compared to the default, but I didn’t expect the effect sizes to all be smaller (and credible intervals narrower) when using the informed prior. Does anybody know why this might be?
So where is the video link Carlos III university?
Wow so prior distribution is data independent/purely parametric
thank you! :)
Awesome
Omg 14:08 INSIGHT
Dirichlet sounds pompous given he (the moustache gives it away) has four christian names.
Who is Zed? :-D
overthinking a coin toss discounts fun some what; more interested in conjugate rites...
This channel deserves more view u ppl!!!!
FREQUENTIST Confidence Interval
Repeated sampling. Each contain N data. 95% of all samples contain value X within a specific interval.
BAYESIAN Credible Interval
95% chance that X value is within a specific interval.
Bayesian*: granted "this" prior distribution and "this" likelihood function.
So it's not that beautiful of an answer either. Though frequentist CIs also require quite some assumptions
1) Find the posterior dist. of B/y when sigma square is known and B is unknown, using uniform and jaffreys prior
Plz help ma before 12am
Por alguna extraña razón las mujeres le dan más valor a lo que escuchan que a lo que esta escrito.
Y valgan verdades, la sociedad es femenina por definición
Por alguna extraña razón la gente le da más valor a las promesas electorales de aquel que se autoproclama defensor de tus derechos y te pide el voto a cambio de mejorarte la vida.
Sin embargo, luego de unos años, las únicas vidas que han mejorado son las vidas de aquellos que se habían autoproclamado.
Si la política consiste en engañar a la gente y a cuanta más personas engañes mejor.
Pregunto, ¿tu te sientes parte del problema o de la solución?
- IMV P2rtido Polític0 en WordPr3ss
-- El primer partido político con funcionamiento interno verdaderamente democrático en la historia de España.
-- Sin machitos alfa, sin capataces a lomo de caballo blanco, sin tonto pollas, sin chulo playas, sin cuadros, sin mi3rdas