Why Do We Need to Perform Feature Scaling?

Рет қаралды 133,081

5 жыл бұрын

Hello All,
In this video we will be understanding why do we need to perform Feature Scaling. Happy Learning!!
amazon url: www.amazon.in/Hands-Python-Fi...
Buy the Best book of Machine Learning, Deep Learning with python sklearn and tensorflow from below
amazon url:
www.amazon.in/Hands-Machine-L...
Connect with me here:
Twitter: / krishnaik06
Facebook: / krishnaik06
instagram: / krishnaik06
Subscribe my unboxing Channel
/ @krishnaikhindi
Below are the various playlist created on ML,Data Science and Deep Learning. Please subscribe and support the channel. Happy Learning!
Deep Learning Playlist: • Tutorial 1- Introducti...
Data Science Projects playlist: • Generative Adversarial...
NLP playlist: • Natural Language Proce...
Statistics Playlist: • Population vs Sample i...
Feature Engineering playlist: • Feature Engineering in...
Computer Vision playlist: • OpenCV Installation | ...
Data Science Interview Question playlist: • Complete Life Cycle of...
You can buy my book on Finance with Machine Learning and Deep Learning from the below url
amazon url: www.amazon.in/Hands-Python-Fi...
🙏🙏🙏🙏🙏🙏🙏🙏
YOU JUST NEED TO DO
3 THINGS to support my channel
LIKE
SHARE
&
SUBSCRIBE
TO MY KZfaq CHANNEL

Пікірлер: 136

@Official-tk3nc 4 жыл бұрын

If you are watching this in lockdown believe me you are one of the rare species on the earth who are working hard to achieve something in their life. Many students are wasting their time watching youtube , playing games, pubg, watching webseries, netflix, etc, all the best, nitjstudenthere.

@shoaibsh2872 4 жыл бұрын

Thanx bro for that comment :D

@justanaverageguy4739 4 жыл бұрын

thanku bro i got today's motivation

@nabiltech1366 3 жыл бұрын

Yup bro.

@masthanjinostra2981 3 жыл бұрын

It’s too late

@makrandrastogi5588 3 жыл бұрын

@@masthanjinostra2981 its never too late bro !!

@ulysses_grant 4 жыл бұрын

May God reward you in centuple for your work sir. We can see you are really know what you are teaching, and that you are passionate about teaching it. Thank you.

@nyinyisoe3604 3 жыл бұрын

This is a very clear explanation of what really scaling is. Thanks for uploading this video.

@siddharthdhingra1 5 жыл бұрын

I wait for your videos every day... You explain so well. Keep up the good work! :)

@yingjiedong3568 2 жыл бұрын

Wow, you are a amazing instructor of Machine learning. I am so lucky to find your videos. Your explanation is clear and easy to understand. I can tell you really understand and have experience as well. Thank You!

@jithinkgeorge2237 5 жыл бұрын

Started watching your channel from last week, Such a great videos.

@qamilmirza4739 3 жыл бұрын

This is amazing content man! Decided to learn ML during lockdown and your videos have been great in explaining the technical aspects when it comes to data. Keep it up!

@geraldlivingstone8104 4 жыл бұрын

Krish - Great job and thank you for posting, better than my MBA professor presented, I now understand.

@mrrasel871 4 жыл бұрын

Great! I want to say I am enjoying your each and every tutorial not only learning.

@zulfiqarali-zq1rg 4 жыл бұрын

I got something new is your every video that is very important for my project .thx

@tirumaleshn8504 4 жыл бұрын

Krish Naik sir! You were amaze with this approach of teaching. A lot better than online courses. And sir, pls arrange videos of playlist in sequence.

@RahulDable 3 жыл бұрын

Thanks for highlighting SGD here, never thought in that way. Great Intuition.

@Lucia-el6ex Жыл бұрын

Gracias por compartir todos tus conocimientos con la comunidad! Eres un crack!

@_HrickKarnaRoy 4 жыл бұрын

Amaizing vid brother just impressed.... Nice way to explain..

@YashSharma-eq3sn 10 ай бұрын

This was helpful. Easy and quick answer to the question beautifully explained.

@eleanortay7351 11 ай бұрын

Grateful for all the knowledge you have shared. It makes my learning journey much more interesting and easier to remember the key concepts now. 🙏🙏🙏😊

@fahdciwan8709 4 жыл бұрын

thanks brother!! found this very helpful

@m0tivati0n71 Жыл бұрын

Wow, very succinct and simple explanation. Thank you.

@deeplearningwithjay 2 жыл бұрын

Thank You Sir , The Scaling makes so much sense now

@benvelloor 4 жыл бұрын

You had mentioned that Feature scaling helps speed up algorithms which use gradient descent. Even though Xgboost is an ensemble learning algorithm, it uses gradient descent to calculate the loss of each model before passing on the wrongly calculated data points into the next model. Isn't it advised to use feature scaling for xgboost then?

@akhileshkumar-mu8gb Жыл бұрын

your classes are really helpful in understanding the basics of ML.

@dontgiveup5271 2 жыл бұрын

Nice!! Simple and clear explanation

@jamesclicks1344 2 жыл бұрын

That make sense! Thanks. It's like using "log" when a dataset has a very high standard deviation

@saqibkhanbhai1859 4 жыл бұрын

Helpful explanation. Thanks.

@tanaygupta2865 3 жыл бұрын

Nicely explained. :) Thanks K. NAIK. :)

@wildananugrah 3 жыл бұрын

thank you for the video, it has enlightened me

@Gaaravify 2 жыл бұрын

amazing explanation, thank u bro

@Person-hb3dv 2 жыл бұрын

explained way better than so many paid courses

@dkhundley 3 жыл бұрын

Excellent video!

@tempura_edward4330 4 жыл бұрын

great video! thank you !

@amilachathuranga5541 2 жыл бұрын

thank you..understood well.

@atoffice7323 3 жыл бұрын

very good bro.. amazing

@windaseptarini9647 3 жыл бұрын

Thank you for clear explanation

@AllRound8255 Жыл бұрын

This kind of content😍

@pranjalgupta9427 4 жыл бұрын

Nice explaination

@himalayasinghsheoran1255 4 жыл бұрын

Thanks for the video :)

@abhishekshrivastav6193 4 жыл бұрын

awesome sir....!!!!

@abirbiba3307 4 ай бұрын

Thank you so much ❤

@atiladursun7075 3 жыл бұрын

Good explanation man thanks

@leoyuanluo 3 жыл бұрын

voice of reason!

@rpnsamuelaigbornajohn9096 5 ай бұрын

thanks alot for this now i know when to use scaling and when not to use scaling

@md.khaledhasan8191 3 жыл бұрын

Well explained

@MrKB_SSJ2 2 жыл бұрын

Thank You...

@Pride_Of_Ultras 2 жыл бұрын

Love the way u speak :D

@ladan1360 4 жыл бұрын

Thank you 👌

@AManKumar-pz6et 5 жыл бұрын

Making video might be of 3 reason 1. You want to learn 2. You want other to learn 3 both 1 and 2 If you want 1 then keep continue Else please make a systematic video or make a website like where any one can learn you can also make it paid if you want but in minimum cost like 2999 or 3999 rs. This would be very helpful for you and others.. Its just an advice. Thank you.

@krishnaik06 5 жыл бұрын

Sure I am planning to start some online courses with a minimum amount

@uttamgupta5653 Жыл бұрын

Sir mere liye toh aap hi Andrew NG sir ho thanks sir for this all content

@vishalkap62 5 жыл бұрын

Do we need to scale our dependent variable Y, as well...

@khushalivaghani61 5 жыл бұрын

Hello sir, All the videos related to ML posted by you are really helpful. Your teaching methodology is also very nice and I can learn easily. Can you make a video related to Audio Processing/ Speaker Diarization? As I want to dive deep in Speaker separation from the audio recording. It might be useful to others also. Thank you.

@alon.hartman Ай бұрын

thank very muhn blud really aprewcite it fromi srael !

@lijinhui6902 4 жыл бұрын

thx so much

@rudrapavankumar4607 4 жыл бұрын

super sir

@sandipansarkar9211 2 жыл бұрын

finished watching

@abhishekdubey2328 3 жыл бұрын

we can use feature scaling on SVR ? it is recommended or not.

@0SIGMA 3 жыл бұрын

if there are classification problems, we should do encoding first and feature scale over it ?

@MasterofPlay7 4 жыл бұрын

what does linear regression have to do with gradient descent? I think the purpose of standardization in linear regression is cuz to reduce the effect of extreme values/outliers have on the model

@aryansinha1564 2 жыл бұрын

Parameter optimization for linear regression is done by gradient descent

@NadinMagdy 4 жыл бұрын

I'm still not sure why it is bad to have a big range of numbers, scaling for the normal number or if feature scaled between 0 and 1, will end up giving the same curve shape if the axis scales are adjusted, can someone help me undertsand

@deepaktiwari54 5 жыл бұрын

Sir, please make video on github, how to use it, how it is help to make profile stronger.

@muhammadshariqkhan6065 3 жыл бұрын

I have trained my model on international data based on Decision tree algorithm but when it comes to test set that is based upon pakistani data . So the answers are not coming out to be same because the data set on which the model was trained with the features was not having mentioned any units and in my test data , I have the units as well. So how feature scaling is not important for this particular case i.e. linked with Decision tree algorithm ? do reply

@sadiasafdar4788 2 жыл бұрын

hello sir, perfect explaination. please can you make vodeo , how to perform CNN on matlab using image dataset..... i will be greatfull to u for this.

@vrushankkhadgi9118 3 жыл бұрын

Nice jacket sir

@apoorvasrini2196 2 жыл бұрын

Doesn't XGBoost use Gradient descent to find the the minimum of loss function? Shouldn't we use scaling in that case?

@mohammadarif8057 4 жыл бұрын

Can you specify from which of your playlist is this video is from

@SahilShah-cd5bi 2 жыл бұрын

Do we need feature scaling for dataset containing values only 1,0, -1 only

@palavirajgude5717 4 жыл бұрын

you talked less about why scaling is important, i didnt get it, practical examples would have been better

@sujanpoojary5284 3 жыл бұрын

i think in his stats video on snd he explained

@PuneethSaiBhaskar 4 жыл бұрын

pleade do make a video on how to select features only required for the dependent variable . (when there are hell lot of variables/features ) :)

@appliedskill 3 жыл бұрын

Something is known as forward selection strategies. You can apply this idea

@arcsaber1127 2 жыл бұрын

So basically perform scaling wherever gradient descent or euclidian distance is used?

@GelsYT 4 жыл бұрын

Thank you so much sir Krish! You explained it very well thank you! because of you I clearly understand it now THANK YOU!!

@Letsbetog 3 жыл бұрын

can i use feature scaling in Polynomial regression?

@mercyjhansi8190 2 жыл бұрын

Is scaling necessary for LinearRegression (OLS)?

@hariprasad1744 4 жыл бұрын

Hi krish, I am unable fill the google form, it is no longer available. Can you please help me

@ankitganeshpurkar 3 жыл бұрын

Should be we perform feature scaling in SVM, Naive Bayes, logistic regression and stochastic gradient descent??

@ankitganeshpurkar 3 жыл бұрын

If yes, then which method should be preferred

@manavshah2119 3 жыл бұрын

Sir please can you given some practical video on this on real data sets

@nihalshukla7718 3 жыл бұрын

Hello sir can you organize your statistics and feature engineering playlist. I think it is not in sequence.

@RahulKumar-lv9yz 3 жыл бұрын

Did anyone get the feature engineering material that Krish keeps on mentioning? If yes, can you share it?

@adwaitpande11 4 жыл бұрын

Is there a scenario where feature scaling may adversely affect the algorithm?

@souravbiswas6892 3 жыл бұрын

@appliedskill 3 жыл бұрын

Think about salary in two different currency. Does feature scaling in this case??

@dimitrisproios1860 3 жыл бұрын

@@souravbiswas6892 is there a proof?

@souravbiswas6892 3 жыл бұрын

@@appliedskill absolutely

@souravbiswas6892 3 жыл бұрын

@@dimitrisproios1860 build the model without applying feature scaling, you will get the proof 😄

@cutyoopsmoments2800 5 жыл бұрын

Bro, is features engineering linked to the machine learning???

@zenilroy1793 4 жыл бұрын

Yes it is part of data preprocessing

@nandhukumar86 3 жыл бұрын

How about Naive Bayes, it does not deal with any distance but the probability. If my understanding is correct NB doesn't need Feature scaling?

@appliedskill 3 жыл бұрын

Yes, You are correct.

@abhishekshrivastav6193 4 жыл бұрын

Sir I didn't get what actually fit and transform do.

@aguyfromparalleluniverse 2 жыл бұрын

thats how you earn a subscriber

@nirajsharma7384 4 жыл бұрын

Sir their is no video for feature scaling

@thelastcipher9135 3 жыл бұрын

Why does scaling not skew the data? I understand why it makes the algorithm more efficient, but I feel like scaling skews the results. Why does it not seem to be the case?

@ashishdwivedi575 5 жыл бұрын

Sir who have an upper edge in data science a cs btech or stats undergrad student

@mashalnabh2747 4 жыл бұрын

B Tech, even if its Mechanical :)

@ayushmangupta6682 Жыл бұрын

I am seeing this video and I want the material but google form is closed so if possible can you share the material please

@siddhantagarwal431 5 жыл бұрын

Sir can we use java for AI and ML (like weka and deeplearning4j frameworks)

@krishnaik06 5 жыл бұрын

Yes u can definitely

@siddhantagarwal431 5 жыл бұрын

So sir why not companies using java over python because it us faster than python , highly portable , better concurrency etc

@chakrapani_nallam 4 жыл бұрын

@@siddhantagarwal431 Python can do lot..more than just few machine learning libraries..it can do preprocessing, visualization and you almost have libraries to work on anything.. it's very.powerful and writing code is easy!

@tirumaleshn8504 4 жыл бұрын

@@siddhantagarwal431 @Siddhant agarwal i think all there previous work were done in other programming languages such as java and all. But when python entered, they may feel difficult to convert them to python.

@adiflorense1477 4 жыл бұрын

7:32 sir, what about naive bayes

@appliedskill 3 жыл бұрын

It is classification algorithm (Idea came from Baye's theorem). It works very well in text or spam classification problem. For more information you can read others blogs on internet

@letslearndatasciencetogeth479 4 жыл бұрын

can someone please share the materials

@ninjawarrior_1602 4 жыл бұрын

i still have doubt that how does scaling helps because even after scaling the relative distance doesnt change

@shankar3109 4 жыл бұрын

Computing gets faster

@ninjawarrior_1602 4 жыл бұрын

@@shankar3109 ok thanx mate

@kumarrishu3026 5 жыл бұрын

Sir can language like c++ or java can get us jobs

@krishnaik06 5 жыл бұрын

Yes they can...

@nabiltech1366 3 жыл бұрын

My fav netflix series

@SP-jv5lv 5 жыл бұрын

Hiii bro.. Im passout this year in entc stream. I don't have knowledg of programming. Im going to learn java directly... Can i go direct or learn first c or c++?? Please answer 🙏 Then im going to machine learning.

@himanshubhandari1783 4 жыл бұрын

Go direct.. go for python

@husnaramohammed8881 3 жыл бұрын

sir iam ur new subscriber can u please share the material with me

@Nobin12346 3 жыл бұрын

@devendunegi247 3 жыл бұрын

your intent is good but i would say try to explain in such a way that a layman would understand what you are trying to tell...like you did not explain 'why' a lot you just told good things about feature scaling but you did not put time in explaining why is it used..

@reachDeepNeuron 4 жыл бұрын

It dint come out clearly

@lakshmisri8978 3 жыл бұрын

What about logistic regression?

@shubhamchoudhary5461 3 жыл бұрын

Yes. Logistics regression also requires feature scaling...

@lakshmisri8978 3 жыл бұрын

@@shubhamchoudhary5461 Okay thank you! Great video btw!

@vineethtangedudona9672 Жыл бұрын

@@lakshmisri8978 logistic regression can be trained without scaling .

@its_mrpanchal 3 жыл бұрын

kuch bhi samajh nhi aaya sirji, aap to kuch or topics pr hi bol rhe ho,,,, aap simple ye btaao ki scaling kyon karni chahiye

@vinaykumar15 3 жыл бұрын

Its very confusing video and doesn't answer why do we need feature scaling.

@suyashneelambugg 4 жыл бұрын

This is what happens when you learn from practical implementation. You have no real mathematical answers to anything. More like somebody learn to press buttons and was able to make some sense of the process. Stop calling Eucledian distance as Eucledian distance unless you can explain people why euclediam norm (L2 norm) is preferred over L1 norm which gives us the Manhattan Distance....

@krishnaik06 4 жыл бұрын

U better check my eucledian and Manhattan distance video ;) you will get to know about the norms.Probably you are taking a conclusion befor seeing all the videos. :-)