Love the work you do for the community ... hence decided to be your Patreon ... this is a big deal for me ... since I did it for the first time. Keep up the good work!
Absolutely great video!! I love this Code With Me format, you can see the theoretical points through practical examples. Best and most natural way to learn!!
I really appreciate your lucid explanations in all your presentations on this subject. In this video you did touch upon the of mean and standard deviation values for the sampled conditional distributions , asking the viewer to see earlier videos for explanation. Perhaps it would improve the presentation, if you at least mentioned that it is due to the bivariate dustribution's correlation parameter (rho = 0.5).
I just discovered your channel. Your videos are excellent. They have a good balance of math, intuition and coding. Thank you very much for sharing your knowledge with us. As a suggestion, the audio of the videos can be improved. It's very low.
I also did not really understand why he showed the correlations? However, it is good to check the autocorrelation in case of Gibbs sampling. If there is high autocorrelation in the chain then each step is possibly not added enough information and this could be inefficient as it would take a looong to converge. I don't think he spoke about it in his videos on gibbs sampling .
Love the work you do for the community ... hence decided to be your Patreon ... this is a big deal for me ... since I did it for the first time. Keep up the good work!
Wow thank you! I appreciate all your support and especially the constructive comments you provide!
@@ritvikmath thanks ... will keep them coming.
Your videos helped me so much during my grad school courses in the previous semesters.
Thanks so much dude
Amazing work! I will definitely support you in whatever little way I can
Absolutely great video!! I love this Code With Me format, you can see the theoretical points through practical examples. Best and most natural way to learn!!
I really appreciate your lucid explanations in all your presentations on this subject. In this video you did touch upon the of mean and standard deviation values for the sampled conditional distributions , asking the viewer to see earlier videos for explanation. Perhaps it would improve the presentation, if you at least mentioned that it is due to the bivariate dustribution's correlation parameter (rho = 0.5).
Great video, really helped clarify the theoretical video.
You are incredible!
Extremely excellent video series! I now start ow wonder how does the Numpy library generate the uncorrelated samples?
I just discovered your channel. Your videos are excellent. They have a good balance of math, intuition and coding. Thank you very much for sharing your knowledge with us. As a suggestion, the audio of the videos can be improved. It's very low.
Can you do a video on bijectors.
@ritvik are you a TA at UCLA?
Great vid, please Code for metropolis hasting and hamiltonian monte carlo as well
Can you code gibbs sampler for beta-binomial distrition? :(((
Big fan of the code along
To use this code do I need to install matplotlib?
well for me in gibbs sampling we don't want to deal with some crazy marginal like 100+ integral
so niiiiiice!!!!
Thankyou sir.💙
Awesome, thanks for this.
But why is such correlation a problem, if that is what you are implying?
I also did not really understand why he showed the correlations? However, it is good to check the autocorrelation in case of Gibbs sampling. If there is high autocorrelation in the chain then each step is possibly not added enough information and this could be inefficient as it would take a looong to converge. I don't think he spoke about it in his videos on gibbs sampling .