Why Lasso Regression creates sparsity?

Ever wondered why Lasso Regression tends to create sparsity in your data? Join us in this exploration where we understand the concept in detail.
============================
Do you want to learn from me?
Check my affordable mentorship program at : learnwith.camp...
============================
📱 Grow with us:
CampusX' LinkedIn: / campusx-official
CampusX on Instagram for daily tips: / campusx.official
My LinkedIn: / nitish-singh-03412789
Discord: / discord
E-mail us at [e-mail]

[@ashishshejwal8514]

I don't know why but people understand most things via KZbin.And reason for this is only this personalities.Thanks sir

[@visheshmp]

this is what I was trying to understand. You are one of the most amazing teacher I have ever seen in my life.

[@barshabanik7212]

here you solved my doubt at 2 in the night...thanks

[@balrajprajesh6473]

God level teaching!

[@vipulritwik]

Thank u for such an awesome Machine Learning playlist. The way you have explained all the concepts is commendable

[@vaibhavchaudhary5571]

The best explanation available on internet

[@barunkaushik7015]

Just an amazing learning experience. Thank you.

[@pravinshende.DataScientist]

best explanation as ever ... from your vedios I getting ans . for every why and this is I like it... thank you sir very much!!!

[@Tusharchitrakar]

You are the best ML teacher hands down. I have a fundamental question: The loss function in lasso regression is non differentiable, so i cannot feed it to the gradient descent algorithm, what are the alternative optimization techniques in such cases? Other example would also be the SVM classifier where the hinge loss term is non differentiable at 0. Although you showed the three cases for different categories of m and obtained 3 equations, do we need to programmatically apply this to the gradient descent algorithm using an if condition?

[@procrastinateadda9097]

Can u help me with the explanation of why this would occur in higher dimension? Like when there are d weights to consider about. Atleast can u direct to a source?

[@lvilligosalvs2708]

Speechless. Thank you , Sir!!!

[@a1x45h]

Brilliant explanation, as always! :)

[@ParthivShah]

Thank You Sir.

[@DharmendraKumar-DS]

Great explanation. Why didn't we create our own Lasso regression class?

[@BoredToDeath2357]

1:08 Why does the coefficient value of BMI increase steadily by increasing the alpha value from 0 to 0.01? It should be decreasing, right?

[@divyanshchaudhary7063]

Why put 2 before absolute coeff ? there must be a reason na (without any reason maths me kuch bhi nhi put kar sakte )

[@SatyaManikantaPotti]

10 interviews!! Thank you sir

[@kavyasharma4738]

kya mast samjahya hain (writing in hindi as it's coming from my heart emotions)

[@theartzgallery8]

Haha Agreed😁

[@Adarshhb767]

18:50 sir but if will get zero after increasing the lamda value but since the condition is for only m>0 then if it was supposed to be 0 then it would transfer to second condition which is for m = 0 which removes the lamda value itself , can u clarify

[@jamitkumar7251]

i think ur right, and since for m=0, L does not depend on m, means that L is not differentiable wrt to 'm', therefore the algo stops at m = 0

[@umasharma6119]

Thanks bro ❤️ Your knowledge and teaching style 🔥 🔥

[@jamalshah3657]

can you provide us the OneNote Notes for 100 days of ML?

[@tusharkhairnar1807]

sir the notes u r writing can u plz share these all notes that would be really great

[@heetbhatt4511]

Thank you sir

[@studology67]

@13:10 when m=0 then why are we calculating m

[@alphaghost4330]

there's no case for m = 0, if m = 0 then input variables would only depend on b

[@deepak_kori]

best..

[@kindaeasy9797]

aaarre par lambda 100 pr m=0 and lambda 150 pr m=-5 ???? 21:42 ,,,,as lambda increases m shrinks towards 0 , and lasso mai toh 0 hi rhna chahiye na , -5 kese hua?????????????

[@sushantmutnale]

sir that pdf of explaination

[@muhammadahsan1448]

here at 12:44 |m| = -m?? How this is possible? Kindly explain

[@pritampohankar7137]

I also have the same doubt

[@satishdubey2626]

mod always gives us positive value so if m is negative then |m| should be positive, for example if m= -2 then |m| = -(-2) = 2 (positive)

[@jpatel2002]

He meant in terms of derivatives, we had put |m| = m in derivatives for m>0, so for m

[@bhartichambyal6554]

please also post the lasso regression python code from the scratch for 3D data

[@canopus4]

So in ridge regression, if lembda approached infinity then m should be zero na?

[@wewesusui9528]

Nope then the coefficient will be 0.00000♾️

[@ManasMohan-h3n]

m less than 0 , than how come mod m is -m it should be +m, Maybe I am wrong -can anyone explain?

[@ndbweurt34485]

It is a very common doubt... Let us we have a number x, we don't know if it's positive.. negative or zero... Now if it was positive..then after applying mod function it will remain same. Example if it was 2. Then 2 mod is 2 same as 2. If x was negative then after applying mod, sign will get reversed Example if it was -3, after applying mod it will become +3..reverse of what we had.

[@alphaghost4330]

@@ndbweurt34485 we simply break the mod into three cases x > 0, x = 0 and x < 0, whenever there's need for differentiation!

[@gauravpaithane]

🤩🤩🤩🤩🤩

[@sayandeepsarkarju_me7821]

Gawdddd