Tutorial 35- Logistic Regression Indepth Intuition- Part 1| Data Science

[@kiranupreti5619]

i really can't describe how many platforms i went through like statquest, datacamp, geeks for geeks, stanford, udemy ,towards data science and much more but never saw such explaination . this is the first time when i am commenting on a youtube video .your work made me do so .you are seriously a great teacher krish . thanku for sharing your knowledge in such a clearer way.

[@mickolesmana5899]

Yep coming from StatQuest.....

[@RhythmShakti]

Me too

[@dharamveer4899]

I think it is even better than 3blue1brown's explanation. Don't you think so?

[@AkshayRakate]

try applied Ai course

[@Nice_lolat]

Other sites just discus about the code. I believe a ML engineer should have a good knowledge of some aspect of maths and statistics

[@ROHITKUMAR-gq8bz]

its really wow!! thank you so much sir for this beautiful explanation.

[@siddharthapothukuchi2687]

Best explanation, better than my prof😂

[@ltoco4415]

At 08:41 you mentioned that because of outlier our best fit line has changed and we are not getting desired output so we should not use linear regression. But a person would say if we get rid of such outliers we can still go ahead using Linear regression and ideally we should be removing outliers. So how do we support the fact that we should not use Linear regression?

[@ouryly1541]

He said there are two reasons: the outliers and the outputs>1 or outputs 1 or outputs

[@dyfarms9441]

@@ouryly1541 Greater than 1 means it is even greater than 0.5 .. same with negative... Not sure of ur explanation

[@dyfarms9441]

What if your outlier points are the input and you need to predict the results for it, Please refer the video in this link kzbin.info/www/bejne/aIXZfI2kia12aq8

[@Tracks777]

nice video

[@snehalvaidya5843]

Thank you 😊

[@codewithedison2199]

very good

[@sharathkumars9135]

Greater than 1 means it is even greater than 0.5 .. same with negative... Not sure of ur explanation

[@pavan9749]

krish how many adds in your vid please manage this..... thanks

[@RajaSekharGowda]

Sir, you are my role model... Your aim " sharing knowledge " is incredible... #Keep going

[@gitadanesh7496]

Hi Krish. Thank you very much for your videos. I just have questions regarding the reasons you mentioned why we cannot use linear regression for classification. 1) outliers : in general outliers make the model inaccurate and we should fix or remove them. 2) what if y>1 or y 1 and and y

[@saptarshibandopadhyay5902]

I was searching the comment section for answers to this question, but no one even questioned this

[@kirubababu7127]

I have a doubt, We are not passing a raw data to train a model, Anyhow we do preprocessing, from That we can remove outliers.

[@nitinverma8585]

I doubt your reasoning... Shouldn't we remove outliers first before fitting any model? If we have outliers even in regression problems, won't the results be bad even for linear regression because the best fit line will be wrong?

[@PetersonCharlesMONSTAH]

Hey bro, can you help me out with from sklearn.linear_model import LinearRegression? I get an error message from using. This is the error message I'm getting. from R and python.ModuleNotFoundError Traceback (most recent call last) in () ----> 1 from sklearn.Linear_model import LinearRegression 2 regressor = LinearRegression() 3 regressor.fit(X_train, y_train) ModuleNotFoundError: No module named 'sklearn.Linear_model' Thanks bro.

[@DS_AIML]

Use from sklearn.linear_model import LinearRegression .You are using Capital Letter in linear_model syntax.

[@ganeshrao405]

You're one of the reason for me to dive in data science community. I learned from your videos and interesting , felt more to learn. Thank you Krish.

[@pankajsoni3215]

what's your position now ? please reply I am a freshie to data science from nit

[@harivgl]

What is the problem with the weight intercept on the linear regression line being greater than 1 or less than 0 ? We're classifying only based on one criteria of the intercept being greater than or less than a particular value like 0.5, right?

[@yashodhansatellite1]

Amazing Krish.. thanks for such effort to explain it so easily. You are our Andrew ng

[@rishavpaudel7591]

Sir I love when you say the word-------> "UNDERSTAND" :p :p

[@bolthirani9504]

listen between 3:05 to 3:15 with deep learning (deep hearing) there is fun

[@dhrubhajong8133]

your channel has brought a huge transition in our lives , especially people who are looking for a career transition. Thank you and your channel for sharing your knowledge and making it easy for us .

[@PREMKUMARBORUGADDA]

tutorial 34 was clarified my doubts, I am waiting for part 2 & 3 of performance metrics for Classification problem in ML. Please upload the part 2 & 3 of tutorial 34 sir.

[@sandipansarkar9211]

Superb explanation.Earlier i was always confused about logistic regression and linear regression as I was unable to understand the mathematical intuition behind it.This is mainly asked as interview questions in product based companies I guess. Thanks Krish.

[@shatadruroychowdhury6319]

I can't find the exact playlist where this video belongs. Can u or anyone pls help?

[@praveenbhatt3127]

kzbin.info/aero/PLZoTAELRMXVPBTrWtJkn3wWQxZkmTXGwe

[@kiranchowdary8100]

Thank you bro please continue it helps a alot

[@Tracks777]

amazing stuff

[@ManishKumar-qs1fm]

Awesome explanation Sir, m waiting ur video on daily basis, sir plz upload maths behind every algorithm, which is helpful for us

[@kirandeepmarala5541]

Saying Thanks is not enough to KRISH Sir...Daily, Your Video make's my Day fullfillment.....Praying God to have grace over you all the time...

[@roshankumargupta3711]

Great video. Can you also help me with the difference between Multiclass and Multilabel classification?

[@anandhuded3647]

Best ever explanation thanku so much for your effort👏👏👏

[@punitkatiyar8564]

Hi Krish, I loved the way you are teaching us about data science concept. I saw the above vedio and have one doubt in it i.e. If we remove the outliers from data, can we use linear regression for classification problem?

[@vinodkumarreddy7696]

Thank u so much sir... Very good info 👍👍

[@DappuDon]

why are we training on this data set this data set doesn't follow the assumption of linear regression.

[@giornogiovanna4954]

Sir, I have a doubt, what if I remove the outlier? What more could I tell to satisfy the interviewer?

[@equiwave80]

Simple concept but the explanation you gave was quite powerful so that we will remember the core concept... Thanks Krish 👍

[@harshitlamba155]

Hi Krish, this is well explained. However, why are we even considering 'adding' an outlier to existing model, which would change the threshold? If we just predict using 0.5 threshold for the new sample point without including it in model, it will give correct answer. By the logic explained in the video, every model will change the hyper plane on addition of new data points. Please let me know what f I am missing something.

[@Sajjad4739]

Hy, my dataset containing 297 features and 9 types of prediction and results with Logistic regressions are low, why is it not a binary formate outcome so results are poor????

[@sumangalbhattacharya6022]

why we are assuming 0.5 for the first case? if the vertical line passes through (75,0) intersect the line at (75, 0.5), then it is okey..... but for the modified line after including new points .. we can thing if it is greater then 0.3 then obese and less than 0.3 then not obese .... so whats the problem ??

[@kushswaroop7436]

At 6:55 to 7:10 you have highlighted point as negative why? Probability cannot be below 0, I could not understand that particular point

[@praveenbhatt3127]

#doubt 9:10, just like we indirectly took 0.5 as the threshold point for determining obese and not obese, and when we had some data point far from other then we are also considering 0.5 as the threshold point, why cannot we also change or update our threshold point according to the data, like we could have updated our threshold point to 0.3 and again it would work fine?

[@lvilligosalvs2708]

This is the best explanation I've seen about this topic. Thank you very much, Sir.

[@vdmaheswari4943]

Great lecture, thankyou so much for everything

[@prajwalwaykos2614]

hey where is the video for explaination of cohen kappa, ROC curve etc metirx for classification can someone help me with that.

[@kanchanadevi6755]

Can anyone suggest completing IBM data science certification will help in getting technical knowledge n get job opportunties

[@shwetharao3894]

Currently pursuing my MS in Data Science and Business analytics, this is the most simple and clear explanation on Logistic regression!! Good work Krish:)

[@abhishekchakrabarty2930]

How come straight line is fitted over here since apart from 1 and 0, there are no other such y points that are plotted...

[@adriandominiquearante3197]

Sir may we know what book did you used in studying this maths behind machine learning?

[@dsklife]

Can you help me with regression models with multi-dimensional data?

[@tiverekarrahul]

Thanks for sharing this Very Very Very useful knowledge.

[@creatorsayanb]

Great video. The voice was not recorded properly in this video. Other than that all is well.

[@filipmilolipinski5299]

I love these videos. You make this puzzling stuff seem so accessible.

[@venkatachaitanyayadla1794]

Could you please give link to this playlist which contains this video

[@gudejyothi6228]

lost connection between tutorial 34 and tutorial 35 please help me

[@amitdebnath2207]

Brother take my gratitude. Huge respect for you.

[@Hitesh-Salgotra]

bro this comment has just came straight from my heart even the andrew ng have not explained with that much clarity even though i know you somewhere learn but your explaination is quite well as i have done ML from 2,3 online courses but i will check whole your playlist again only for the deep understanding of math behind it

[@soham4741]

andrew explained it really well , but he uses more complex language which is harder for us to relate to. I understood andrew really well too after watching his logistic regression videos twice each.

[@moonlight-td8ed]

bro every youtuber learns from andrew ng

[@motunrayoakinsete]

You are a really great teacher. Thank you for sharing this.

[@sunilpatil1923]

Very good vedeo, well explained...

[@Raja-tt4ll]

Very useful video. Thanks Krish.

[@deepaksurya776]

Everyone likes Andrew Ng machine learning tutorials 😋

[@adahal39]

Haha. Searching for this one

[@shabareesharyan]

You are an amazing person. I hope you know that.

[@krishnab6444]

Great Explanation sir, Thank You

[@flse8337]

This is what happens when you really want to teach and its not all about the money. Krish took the time not just to teach the concept but also connect it to reality. Good stuff boss. Hope the quality of your work only increases.

[@manishkatiyar2403]

Thanks

[@nileshmandlik9662]

make a video and reveal your total income sirji

[@akukweanselem2694]

Great stuff keep it up 🤗🤗🤗

[@mansoorbaig9232]

Comprehensive explanation, Part 2 and 3 please...

[@krishnaik06]

Check the complete ML playlist

[@sardorabdirayimov]

Thanks for sharing your another great tutorial

[@Eng_5567]

11:05

[@neelamegammuthuraj8982]

You are amazing ... was having a long run doubt of why to use logistic regression ... fantastic explanation. Great Work .. Much Appreciated.

[@ashanair1602]

Ads between videos interrupt learning

[@1UniverseGames]

What will happen if we use (-1,1) instead of (0,1) in logistic function, what kind of equations it will give? Any video or source to study this?

[@SAN-te3rp]

Can anyone tell where he stays I have to touch his feet for this incredible talent of teaching maths 🙏🙏🙏🙏 and students who are from primary schools ,don't watch this video and dislike that just bcoz you didn't understand

[@venkivtz9961]

I think the main problem is we use probability to find whether it is obese or not, when you use the linear regression, you got probability less than zero and greater than 1, which is not possible, hence we can't use linear regression in the case of these type of classification problems. If i am wrong, please correct me. I appreciate your explanations on your videos

[@siddharudtevaramani1055]

Having watched many videos related to difference b/n LinR and LogR, nobody explained in this way. I think this is what makes stand out. Just fabulous!

[@archanadevi7081]

Hie. Your videos are very informative and easy to follow. Could you please prepare a video for LR for text classification? Couldn't find it in ur playlist as well as couldn't find any other video in KZbin useful for this topic.

[@123XTSK]

Why use logistic instead of linear for binary classification ?-The reason has been well explained.Thanks

[@satyajeet8438]

In binary classification problems....if we use linear regression and there will be much outliers...the best fit line will be deviated and we will not get the desired results....hence we go for logistic regression...but...for a non classification problem...means continuous values....will adding outliers affect the best fit line?

[@yadav-vikas]

how much our results will differ if we apply Sigmoid function to training data then predict for Linear reg VS Logistic Regression ?

[@kafeelbutt]

sir please upload SVM intuition

[@ganeshreddymangunuru5001]

Awesome sir👌... superb explanation 🙏... please continue to upload videos sir.

[@milansood8240]

Wow sir, abhi pura upgrad ka module padha but know chahiye tha logistic ab samagh aya 🤩🤩🤩

[@rajbir_singh0517]

sir one question and need more clarity. what if i treat the outlier in the dataset, what will happen then can i use linear regression model for it. line which will cross the above 1 and 0, if condition is define that above 0.5 classify 1 and below that classify 0, does it not satisfy the define condition. please clarify these doubt.

[@parthpatadiya9197]

this same thing you can do with single if condition in programming language if(value>=0.5): obise else not then why we should do this sirji?

[@pratikbhansali4086]

We r not getting join option on your channel

[@tejaswiruttala2070]

sir pls give me a suggestion to how to assign weights to features based on current situation....means according to real time problem some features may have hight weightage than other features...how can we assign weights

[@rashmidhawan4783]

Your explanation is beyond excellence... After a lots of traveling finally I find a outstanding platform to learn data science.. I am pursuing Ph.D in machine learning.. And this tutorials make my PhD interesting...thanks a lot...

[@arkojyotisen1929]

sir what's the problem if y becomes greater than 1, or less than 0 as we can simply consider the whether y is greater than 0.5 or not, and come to a result?

[@shaikmahammadali8043]

Mind blowing explanation guruji . Happy teachers day

[@akshaykapoor9492]

How is 75kg obese :( hahah Thanks for the clear explanation

[@rnsahu2002]

sir just one question, if i remove my outlier in all my data points, will the linear regression work in this kind of classification ?

[@AbhishekDaszealous]

still you can't

[@rahulsharma5256]

If we remove all outliers. It should work. I don't know, why it will not work?

[@shivprasadkale4809]

I did Data science program with simplilearn and also reffers much yt channels bt ur the guru we need. much simple language and clearer concept thnx

[@NGH8731]

I must say Sir you are genius. What an explanation of logistic regression. It's awesome and so simple that anybody can understand. If you watch this video twice you will never face any problem in future to explain logistic regression.

[@rajdudhane9641]

well here are we geting the future predictions or are we getting the predictions done by this algo so we can recheck it with the test set ? how can we use this or any other alogo to carry out future predctions

[@mayurpardeshi395]

You have uploaded practicle implementation of all the algorithms ???

[@LifeInNL]

How could I find the 1st tutorial form these series?

[@dm3248]

Lectures of krish & Andrew ng course on ml is a nice combo..!! 😁

[@9902152322]

explanied very clearly and in short, keeping going. all the best sir.

[@abhishekdubey9562]

But sir linear regression is regression algorithm..how we can use it for binary classification

[@mvk11574]

this is much easier and simpler to understand than what was taught in paid course ..thankyou so very much .You made life simpler for non ML background guys like me .

[@sivamutyala3611]

Superb explanation on why we should go for Logistic Regression. I don't think we can find this kind of clear explanantion in most of the channels. Thank you for presenting us the concept in easy and beatiful manner.

[@basavaprabhusr9930]

Hello sir, i have been watching your all videos and it is very well explained sir. but can u make some videos on NLP problems sir like the word2vec algorithm sir

[@krishnaik06]

Already created

[@basavaprabhusr9930]

@@krishnaik06 Thank u sit