Lasso Regression

[@rishabhsharma329]

This question was asked in my interview. 7 mins of this video changed my life, 5 years ago! Thank you

[@laIlI129]

First time, I am exploring the meaning of LASSO Regression and I have no confusion after watching this video. Very helpful. Thanks Ritvik Kharkar.

[@kristjan2838]

Took a convex optimization course last year. you explained clearly in 3 videos, what took days of digging previously. Papa Bless

[@junjiema4613]

very helpful! Like the speed you speak

[@michaelfresco2769]

also exceptionally clear

[@glaswasser]

i first thought I had youtube still on 1.5 speed haha

[@qwqsimonade3580]

thanks for the video so much. I've been confused by dumb prof for almost one year, and until I understand ridge and l1l2 penalty just by a video. thank you

[@glaswasser]

finally! Stumbled upon that figure in the ISLR book but did not understand what was going on, you made it clear to me now, thanks!

[@ritvikmath]

Glad I could help!

[@luqiyao5896]

The best introduction of LASSO, very easy to understand! Thanks!

[@Dhruvbala]

solid video. saved my interest in the subject, so thank you very much!

[@perrysellers9198]

Excellent job explaining Ridge and Lasso. Your equations/functions AND visuals close the loop nicely!

[@ariani86]

your videos are extremely helpful and easily to understand the math behind ML! Thanks a ton!

[@meichendong3434]

Very clear explanation of the contour!

[@ahmetcihan8025]

This insane Man. Thank you so much.

[@hcgaron]

You are awesome. Thank you for your passion to teach this topic!

[@krishnapathi572]

Amazing clarity of idea...and perfect speed for the explanations :D

[@robertc2121]

Brilliantly explained - Brandon Foltz love you video series!!!!

[@Michael-uc3fx]

Awesome introduction, thanks! Keep posting videos!

[@SaintRudi85]

Excellent video. Very clear. Thank you.

[@spencerlee5500]

So great! Thank YOU!

[@secondtonone2284]

Thank you so much for sharing this brilliant video! If you can afford it, I hope you cover the unique feature of adaptive lasso (oracle properties) too.

[@deepakravishankar169]

really succinct and to the point. good explanation

[@iVergilchiou]

It's so clearly and very helpful !! Thank you so much !

[@aliteshnizi672]

Falling in love with your videos

[@akhileshpandey8457]

Most perfect Video on this stuff.. Even the pace was something I could keep with :)

[@faithkalos7745]

Very good explanation, thanks a lot !

[@GotUpLateWithMoon]

very helpful, thanks very much Ritvik!

[@mikx55]

Super precise and incredibly helpful!!

[@jianishen5656]

Thank you so much !! the explanation is so good !

[@ronithsinha5702]

Can you please explain again why exactly the co-efficients of the B vector hit the edges of the pyramid in case of Lasso Regression, but they do not hit the circumference in case of Ridge Regression. This is the only concept I am not being able to grasp that how does Lasso lead to elimination of co-efficients, but Ridge only causes shrinkage of co-efficients and not entire deletion.

[@BhuvaneshSrivastava]

Great videos as expected 😊.. Also please find time to make videos on: - A/B testing - Survival Modelling - Type of errors - GBM

[@ritvikmath]

Thanks! And I will look into those suggestions

[@preeyank5]

thanks a lot...God Bless!!

[@aliteshnizi672]

Incredibly good.

[@morumotto]

Thank you!!!

[@rasikai521]

perfectly expplained. Thank you so much

[@urmumsfrend]

thank you!

[@ritvikmath]

Welcome!

[@shakedg2956]

really good explanation!

[@vishnu2avv]

Awesome Video. Thanks a million for upload :-)

[@abomad2011]

good explanation

[@ravivijayk1840]

thx for doing this video, intuitively helpful! couple of questions, 1) In lasso, are resultant coefficients be always positive or zero? 2) do we still interpret coefficients after they get penalized by whatever lamda value we pass?

[@miliyajindal]

I have been learning about data science from the last 6 months but there is no article or no videos that are better than users.

[@chillwinternight]

Thank you! A very helpful video. Please consider making a video on coordinate descent. :)

[@alimuqaibel7619]

Thanks, very informative

[@lluisgasso]

Awesome Job!

[@victorcrspo]

Hello! I have a question related with this video and with the Ridge Regression video. Why shoud not I use these methods if I have one variable ( Y = betha_0 + betha_1*X) ? What would happend if I used one of these methods in that situation? Thank you!

[@mech_builder7998]

this intuitive explanation made lasso regression "click" by me, so a big thanks! Were you inspired / did you get the ideas / diagrams from a book or did you come up with them yourself?

[@batosato]

Hey there, Thanks for all the explanation. Could you make a video on Non-Linear Least Square (NLS) estimator and how is it different from OLS? Thanks

[@jackjiang7617]

great explanation!

[@Chris-is9fm]

Thanks, cheers!

[@skan121]

Brilliant!!

[@sikun7894]

Thanks!

[@robertjonka1238]

outstanding

[@manikandantv3015]

could you please explain how some of the coefficients are becoming ZERO in LASSO? I would like to know the internals.

[@antonisstellas741]

very nice video!

[@kautukkaushik7587]

Thanks for the video. Explaination is really great. But I have a question, what if the curve passes through line between (c,0) and (0,c) and also between (c,0) and (0,-c) , then which point would be better?

[@harminderpuri1243]

that curve would not be the smallest curve .. there will be curves with lower (y - Bs)^2 .. plot it and visualize

[@tomaspablofermandois4690]

Thanks for the Video!

[@pranukvs]

great stuff man, you should put up a course on udacity or something !!

[@manikandantv3015]

if LASSO is for feature selection how it's different from PCA? Pls clarify

[@nicholasdi1529]

Hello! So will C be 25 in the case? (around 5 mins)

[@dodg3r123]

Thank you so much! So how do you come up with a suitable value for c?

[@thelastcipher9135]

How do you pick the constraint c?

[@tonix1993]

beast lecturer !

[@marcofumagalli8147]

Good job !

[@tableauvizwithvineet148]

What is the meaning of green level curves, why are they used ?

[@yxs8495]

excellent

[@qiulanable]

awesome video !!!

[@jererox]

Thanks it really helped.

[@sidk5919]

Awesome!

[@bitadet3935]

GREAT VIDEO! :D

[@Theateist]

Why do the corners get hit a lot more than other points?

[@nikhilnambiar7160]

So finding new beta is done by taking derivative of lasso formula with respect to beta? And subtracting it from old beta?

[@Jack20032008]

thanks. it's helpful

[@phuccoiinkorea3341]

what is its optimization formular?

[@randomforrest9251]

great explanation, but it's a little bit misleading, since we are not regul. our beta0, but beta1..m.

[@ritvikmath]

You have a good point, thank you!

[@joshespinoza8645]

Awesome, so what exactly are the "betas"?

[@thestyxx]

+Josh Espinoza The regression coefficients, in other words, the estimated effects of your parameters.

[@AhmedAbdelrahmanAtbara]

I just don't agree with you in the feature selection argument, if beat comes with many zeros that doesn't mean the model is conducting any feature selection process there, it will automatically ignore the zeros. Probably feature selection is something different.

[@fredrious]

it's very good, but too fast!!!

[@edlarmore5958]

Great explanations. Just wish you would talk a tad bit slower.

[@zhenqiangsu8231]

赞！

[@jianfengxu7889]

beta_0 should not be in the regularization term.

[@akhileshpandey8457]

he is just using it as an example.. he explained that in Ridge video

[@puifais]

This is a great video. I suggest you do NOT touch or move the piece of paper this much. It'll be less distracting and help the audience look at equations and compare the information.

[@youyangcao3837]

Thank you!!!

[@alvaroneuenfeldtjunior9157]

outstanding