Bayesian Regression in R

[@avocadoarmadillo7031]

New sub, enjoying your channel so far!

[@cat-et6dx]

Very well explained. Thanks!

[@NIZI-c7o]

omg❤ this really saved my life!!! Thank you so much!!! can't believe I could use the model right now❤❤❤

[@nimanemati1777]

Hey Spencer, Thanks for the great video! Could you teach us Bayesian model averaging in R? I couldn't find a suitable tutorial video for it. The available videos lack detailed, step-by-step instructions, and they only cover general points. I would appreciate your guidance on Bayesian model averaging. Thank you!🙏🙏

[@Stat_Guy]

Thank you Spencer, for sharing this great video! It will help a lot to understand bayesian regression

[@lebzgold7475]

Just finding your channel now, you're of great help! Thank you so much sir! :)

[@ayeletesema590]

Thank you Spencer for this great piece of work. I would really appreciate it if you could share reference for the theoretical background. I need mathematical expressions to include it in my current work.

[@SpencerPaoHere]

This was pretty dense with information. Note that my prior here was a Normal distribution, and I used arbitrary constants of a = 0.1, b = 0.1, g = 10.. Once you get to the coding demonstration, I highly recommend that you pull up the code + data to follow along. The code can be retrieved here: github.com/SpencerPao/Data_Science/tree/main/Bayesian%20Regression

[@samsonamani8103]

Thanks sir, is there a written document explaining this?

[@user-wr4yl7tx3w]

Not sure how you clustered with K Means? Is it so that you can apply bayesian to different clusters separately?

[@juanete69]

Do we need to use cross-validation in bayesian analysis to control overfitting?

[@SpencerPaoHere]

Nope. They are fundamentally different. (Bayesian has MCMC that comes into play)

[@juanete69]

@@SpencerPaoHere Then bayesian models can't overfit if they have too many parameters or very few observations?

[@SpencerPaoHere]

@@juanete69 it most certainly can! Though the sampling will address the overfitness of the model to the best of its ability via MCMC

[@SpencerPaoHere]

@@juanete69 it most certainly can! Though the sampling will address the overfitness of the model to the best of its ability via MCMC

[@red3987]

How long did it take you to learn to do all this stuff in R and really understand what you are doing? I want to really master this type of stuff using R but do not feel as comfortable with it all (the code and the theory) as I would like to. I am about to get my undergraduate degree in econ and math, then my masters in econ and have been practicing in R for about a year and practicing statistical theory for about 6 months.

[@SpencerPaoHere]

Its a journey! I learned all my basic stuff in college (4 years) and the more advanced stuff while working. I learned R in college and all the python stuff was in industry. 6 months is not that much but around the 1 year mark of continuously using the language: R (or any programming language for that matter) and you’ll be just fine!

[@red3987]

@@SpencerPaoHere thank you !:)

[@simple-jp1cs]

Great video, your explanation is very clear. I am confused in how you got the values in the logit_loan_default column? Does it need to have binary values in this field?

[@SpencerPaoHere]

In the example I used, you can definitely just have categorical variables (0,1). You'd just have to change your distribution to that of Bernouli (2 outputs) The values I have are essentially centroids for each given cluster. (Check out the kmeans_data.r script I have on github). So, in essence, I am doing a regression to fit the normal. Some additional information can be found here: bayesball.github.io/BOOK/bayesian-multiple-regression-and-logistic-models.html#bayesian-logistic-regression I hope that helps!

[@chukwuemekanworu-uo2td]

The video is too blurred, i can't see it clearly.

[@SpencerPaoHere]

Weird. This video goes up to 1080p. You might have to wait for the buffer to go through. (And select the appropriate setting)

[@agracian1]

Hi Spencer, could your share the R code for how you k-mean'd the data to generate Kmeans_scaled_data100.csv?Did you run a logit regression 1st using glm to generate your logit_loan_default field? Thanks Alex

[@SpencerPaoHere]

I uploaded the kmeans code to my github repository with some sample data. (The original dataset is too large). The binary outcomes already came as is, so I did not do any glm related functions.

[@cliffme6310]

can this be used to predict forex markets reversals?

[@SpencerPaoHere]

You could try! Though, I've personally never tried this technique on forex, so I can't comment on how well it'd do.

[@CenasDeWinner]

Since you post your like/dislike count, here's a like from me. The video was great btw

[@cansahin529]

Thank you for your video it so helped to me on my school project appreciate that🙏🏻🙏🏻👏🏻

[@SpencerPaoHere]

I'm glad I helped! :)

[@taotaotan5671]

Nice meme 3:31 lol

[@SpencerPaoHere]

😂

[@wave_chaser_n]

I see one punch man pretty cool

[@SpencerPaoHere]

Ayyyyyy :)