Видео

hey. your work is excellent!!!! where can I get the dataset that you used in this ML project

Great video!

Thank you for the terrific video. I wish I had seen it earlier! These limitations should be fully disclosed in the documentation, even in PyMC 4 I'm facing a bunch of issues when running posterior predictive as soon as the model gets slightly complex

This is a really good presentation. Thank you and congratulations!

very good video and waiting for the next video.

The memes are on point Incredible explanation 10/10

Great talk! :) Is the notebook downloadable somewhere to better follow?

RSVP for the upcoming Q&A session to explore more about it. 1. For AMER regions (29th Jan at 15:00 UTC)👉 www.meetup.com/pymc-online-meetup/events/298279180/ 2. For Asia regions (30th Jan at 06:30 UTC)👉 www.meetup.com/pymc-online-meetup/events/298356399/ More details are here: discourse.pymc.io/t/new-pymcon-talk-enabling-uncertainty-quantification-by-anne-reinarz-linus-seelinger/13583

truly useful and thanks for sharing this! is there any code or use case?

This didactically extraordinary 🎉

Great video! Anyone knows if there is another one of adding additional regressor?

thank you for this tutorial, it's really help me to finish my work!

I want to give a lecture on Bayesian using pymc3

Thank you for sharing your story. For the first time I feel like Iam not alone. Going to read the book.

Thanks for this video. In minute 20:42, can you explain where the normal parameters for "c" come from? My guess is they are only needed for the "do" operator and mu=0, sigma=1 are assumed to be known parameters of the future distribution of tv ad spend. Conjecturing further, if the distribution of tv ad spend was not assumed known, these parameters could be random variables as well. So, when the do-operator is used to get P(Y|c,do(z=0)), does the "c" here refer to the observed tv_spend or the observed output of a Normal(0,1) rv?

So cool!

How did you chose prior distribution

Sorry if this is obvious but what module is `pt` (the one with exp and abs in the 1-to-1 and many-to-1 transform examples)?

an insightful presentation, kudos to you guys!👏👏 Any suggestions on how to built this, a media mix model from scratch using lightweightMMM in python?

Good video. Just a quick question: When we take gradients of the ELBO wrt the variational parameters, can we actually write the MC Approximation given in 10:22 ? gradients of expectation will not be equal to the Expectation of gradients when we're sampling from the same distribution wrt which we're taking gradients. In ADVI, we end up reparametrizing to Standard Normal DIstribution. Any explanation would be helpful

Thanks for the talk. That's maybe not the place to ask questions but it will help your referring: Is this pymc_experimental.marginalize only for discrete variables (which were already very easy to marginalise by hand) or is it able to work with some continuous functions? Automatic marginalisation of continuous variables would be awsome! Which is certainly related to the posterior conjugacy you're discussing in conclusion: those are the cases in which marginalisation is easy.

Thanks for this wonderful support. Sincerely grateful

You and Fossenbeck are the best explainers of GP’s out there! Well done.

Great talk

Excellent talk. I appreciated the commitment to starting simple and iteratively adding complexity. Many resources dive right into complexity without robustly detailing the “why”

This video was very informative for me. thank you.

I love this approach because it really shows the power of Bayesian analysis. Often in the frequentists world ( and the ML world) we are confined to the models that have a relatively "known" structure (regression, ML models with dependent independent variables features, etc). This approach is almost more scientific in that your testing your model of the DGP more explicitly. Its akin to like a physics equation that describes motion or energy rather than a regression model and attempts to do the same. Very cool stuff!

Thanks for sharing! :)

Wonderful talk - delivered like an expert magician. I would recommend it as 1st talk for someone embarking on pymc journey. Great example of value of these methods in 25 minutes. Thank you!

What a wonderful explanation !....Thank you so much.....Is it possible to get an R code for this ?

Excellent !

Amazing Presentation and Explanation. Congratulations

Link to Jin and colleagues, please.

there seems to be some deprecation in the code. for example, with gp = pm.gp.Marginal(mean_func, cov_func) TypeError: Base.__init__() takes 1 positional argument but 3 were given

Really great video! Learned alot about prophet -- general mcmc -- and liked the package development stuff

Finally, pymc4 released