Least squares using matrices | Lecture 26 | Matrix Algebra for Engineers
Вставка
- Опубліковано 8 лип 2018
- Definition of the least-squares problem for fitting a line through noisy data.
Join me on Coursera: imp.i384100.net/mathematics-f...
Lecture notes at www.math.ust.hk/~machas/matrix...
Paperback at www.amazon.com/Matrix-Algebra...
Subscribe to my channel: ua-cam.com/users/jchasnov?...
Find other Matrix Algebra videos in my playlist ua-cam.com/play/PLkZjai-2Jcxlg-Z1roB0pUwFU-P58tvOx.html
This is a great overview of the problem! A teacher of mine went over the steps without all the justification, so this was very helpful!
Great video to quickly review the concepts. Thanks Jeffrey.
Wow. I was initially skeptical when I started watching thinking this is just another video on this topic. I already knew this material but now I know it much better. Very clear explanation. Thank you.
Sir, you are amazing! This is by far the best at least according to me a best explanation for least squares optimization problem.
great video , finally understand clearly the topic
Very helpful, thank you
Ax = b in general has no solution here, since b is in an n-dimensional space and Ax is in a 2-dimensional (at most 2) subspace of that space, since Ax is a linear combination of 2 n-dimensional vectors.
Just for anyone else that got stuck there for a moment.
You are a legend 👍
you are the best!!!
thank you
I think what makes this topic confusing for beginners is that the x vector in Ax=b does not represent the "x's" but rather represents the betas. And the b in Ax=b does not represent the betas but rather the y's. It is an unfortunate accident of the nomenclature.
That would be a beginning beginner.
@@ProfJeffreyChasnov me is a beginning beginner ^o^ ty for the great lectures
great!
Nice sir
it's not necessary that an overdetermined system has no solutions, it is unlikely in a real world scenario but it's not necessarily the case. If the columns of an overdetermined matrix are dependant for example in the 2D case then they lie on a line. my point is that the fact that there is no solution is not because the system is overdetermined but because b happens to not lie in the column space
At 7:17, you said "if AX equals B is overdetermined, that means B is not in the column space of A, there's no solution to X". Where does that come from ?
Is this dude really out here writing backwards
or maybe some tech has been made that can flip a video (omg what has the world come to)...
I think that he is recording himself behind a transparent wall, and then he reflects that recording around the vertical axis.
or he's writing on a mirror and the video might be everything in the mirror
@@cathyfeng3385 writing on glass and then flipping video horizontally
TENET
In 7:15. If Ax=b it is not true that this implies that b does not lie in the collumnspace of A. It may be overdetermined and have a solution.
We are doing the least-squares problem here.