Mod-01 Lec-30 Principal Component Analysis (PCA)

Applied Multivariate Statistical Modeling by Dr J Maiti,Department of Management, IIT Kharagpur.For more details on NPTEL visit nptel.ac.in

[@prajwalshenoy2735]

Highly underrated video on PCA. Explained using a combination of statistical and linear algebra approach. Nothing beats this.

[@gulabpreetsingh9985]

can you explain at 23:57 , how we got z1 = x1cosθ + x2sinθ and z2 = -x1sinθ + x2cosθ.

[@ArthurMcGDM]

Superb explanation. The best I can find on the internet or in the classroom. Thank you.

[@arunbm123]

please tell me how derivate of aj'Saj = 2Saj ?

[@juanfelipeceronuribe2910]

At 31:23 what makes the transformation orthogonal is A'A=I, not inv(A)A=I (that last one holds for any invertible matrix)

[@worldpulse365]

Indian teachers are by far the best. Mainly because many of them have stuck to traditional forms of instruction...patiently explaining and WRITING DOWN everything step by step. Nothing beats this. As a student you don't want to see everything compressed into slides with all the info already there (a sad norm by lazy professors in 'reputable' universities nowadays. You want to see the puzzle unfold as the teacher explains. Slides should be used sparingly as used here.

[@manisekar2884]

dear sir, i completed my eng in 98 after that i lost touch with maths, i joined recently in doct program in managemnet looking for understanding on PCA, the way, you derived the methodolgy of this, rekindled my interest in M1 back.. thank you

[@kennetholelewe9932]

Sir, I love your teaching simplicity. Thank you so much.

[@sumanshekhar3197]

Oh my god! This is amazing explanation of PCA hands down!

[@VALedu11]

Thank you sir for this comprehensive explanation of PCA. You made the derivatives look easy as well.

[@rakeshnalliboyana2160]

Nice explanation before I did not have a proper idea ,but now I got an idea regarding PCA after watching your video.

[@gulabpreetsingh9985]

can someone explain at 23:57 , how we got z1 = x1cosθ + x2sinθ and z2 = -x1sinθ + x2cosθ.

[@vikiviky1122]

He is the answer of why we need a teacher. Thanks

[@kinglucos5146]

It’s always the random Indian guy who explain your toughest subjects like a champ

[@kanchandatta4668]

Sir one thing not clear so far at 23:58 the projetion on z2 why - sin theta is coming? Is it 4 th quadrant so. We know all, sin tan ,cos positive in 4 quadrant. Not sure please response.

[@HelloIamRachnaGupta]

Thank you so much sir. This video answered my questions of PCA implementation in ML which no other tutorial/blog did.

[@prabhuthomas8770]

The best lecture on PCA. Thank you.

[@vaskoo84]

Best explaination on PCA , Thank you sir !

[@ashutoshsingh1808]

What a clear explanation!!! Thank you Sir!

[@vishwajeetsingh5444]

I love the way Maity sir explained the concept of PCA.

[@GIS-Engineer]

What is covariance. What is eigen value and eigen vector.what is correlation. If we have 150 sample of water .what is the relation of and all

[@kweweli7821]

destroyed PCA in an hour...perfect job

[@chinkirai988]

Thanks for sharing beautiful maths of PCA.

[@gouravkarmakar9606]

best explanation of PCA.

[@ashumohanty566]

long live sir.you are the god for poor student

[@Dyslexic_Neuron]

how is the variance of X replaced by covariance of X , didnt get that part

[@ap3035]

Rotation of axis by theta also rotate the data on axis or not?

[@GuilleLissa]

Excellent explanation Dr. J Maiti

[@sathviks1420]

Good lecture! Thank you!

[@kadirgulec591]

Excellent demonstration. But if you have enough background. Thank you.

[@matinhewing1]

Trying to learn the background as I go along...)))

[@extramiles3831]

Sir, can we use Mohr circle here and avoid tedious calculations?

[@jhareswarmaiti8766]

+VIVEK SINGH Please explain a bit the use of Mohr circle here.

[@anigov]

Jhareswar Maiti Thank you so much Sir for the wonderful explanation in both videos.It has given me a much better understanding of PCA --

[@MeghaSKumar]

hi sir..thanks a lot for the clear explanation of PCA. that was really helpful for me. I have a doubt @15.45..dimension of data matrix x is n*p..so n rows and p columns..then why is it p*1 in x in the immediate step..or is that an uppercase p.. final matrix should be m*p where m

[@idntknw1000]

Comprehensive explanation. Thank you

[@dwivedys]

Excellent! Where is the second part of the lecture?

[@shebhk11]

kzbin.info/www/bejne/pKaVk6KDibCCfKM

[@hellomacha4388]

Super explanation

[@amitkumarmaiti5392]

Great explanation

[@jagadeesh2681996]

can you tell me the nam eof the text you are flowing

[@snehalnair3708]

Awesome video

[@arunbm123]

At 37.33 a' V(X) aj .........how did aj come here

[@juanfelipeceronuribe2910]

Also bear in mind that V(X) is the matrix of covariances of {x_1,...,x_p}

[@hemmapermal532]

arunbm we are maximizing the Var(z) and aTa=1.

[@hemmapermal532]

V(aTx)=E[(aTx - aTu)^2]. This is the formula of variance. As we already knew E(aTx)=aTu. u=mu. You try expand the equation and you will know where that aj came from.

[@Dyslexic_Neuron]

@@hemmapermal532 taking aT out , we will get (aT)^2 and E(x-mu)^2 which is Var(x) , finally we will get (aT)^2* Var(X) , but where does the aj come from ??

[@sharathraju8849]

PRINCIPAL COMPONENT ANALYSIS PROCEDURE BY GOMEZ AND GOMEZ...IF ANY BODY KNOW PLEASE REPLY...

[@SAHILGAJBHIYE60]

Ty very much sir ❤

[@jontravolta1625]

Thanks sir!!

[@ridickrolandtakong5358]

Excellent

[@PavanYadiki]

Thank you Sir.

[@deborahawe4063]

Thanks lots

[@diehardcynic]

Thank you sir

[@rojalin9221]

NICE SIR

[@mswoonc]

durka durka

[@sansin-dev]

So much paper wasted. Could have shown everything on slides or board.

[@turtlepedia5149]

These are super old videos of 1990s at that time boards were not so common and thus method is better for understanding dickster

[@ebenezerr9152]

If it was shown on slides, u wouldnt have understood