Linear Regression 1 [Matlab]

This video describes how the singular value decomposition (SVD) can be used for linear regression in Matlab (part 1).
Book Website: databookuw.com
Book PDF: databookuw.com/databook.pdf
These lectures follow Chapter 1 from: "Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control" by Brunton and Kutz
Amazon: www.amazon.com/Data-Driven-Sc...
Brunton Website: eigensteve.com
This video was produced at the University of Washington

[@payman_azari]

I recommend watching the regression sections multiple times; tons of points to learn and to look in the references. Extraordinary tutorials, thanks Steve.

[@hassaanhameed]

i replay the video after reading your comment I need to watch it again after testing on the real data

[@adricat59]

Really useful, thanks!

[@kouider76]

Thank you for the excellent explaination, looking Forward for more Video

[@danielhoven570]

Just got your book! can't wait to dig in

[@Eigensteve]

Awesome! Thank you!

[@x1x2x125]

Great video!

[@user-ym8rz6mw5r]

wow, good job! I take back what i said about you lacking examples. this is top stuff

[@dragoncurveenthusiast]

9:18 I think b = a*x would be more accurate than b = anoisy*x.
After all, it is the true factor a that has an influence on b, not the noisy measurement we took of a.
I think this is the reason why the plot looked so much better than expected, also at 7:58, the noise was added to a before calculating b from a, so the noise didn't influence the relationship between a and b.

[@songurtechnology]

Thank you ❤

[@fahimmumand]

How can one calculate the uncertainty with svd?
Say that we have uncertainty in both a and b, based on the uncertainty the slope and intercept of y should have uncertainty as well, can you please elaborate how to calculate the uncertainty in the slope and intercept.
Thanks in advance.

[@vvviiimmm]

9:18 More like "the danger of mutable variables" :) Thanks for sharing, keep it up

[@HuadongWu]

the first method demon [time 6:45 - 7:19] is problematic explained as following: [U,S,V]=svd(a,'econ') input a doesn't change slope, so [U,S,V] won't change characteristics, xtilde=V*inv(S)*U'*b won't change its distribution from b's characteristics

[@lingnie2809]

just out of curious. Is that a 13'' macbook or 16'' macbook?

[@jakobjonsson1474]

When using the SVD command, you input your array a as a 1d array. This doesn't work in python. Does anyone know how to do this in python with a 1d array?
Thank you

[@jakobjonsson1474]

I understood what the problem was when watching the corresponding video in Python.

[@jeroendebest1711]

x=a\b

[@x1x2x125]

First!

[@younique9710]

After we added some noise to a, should we run SVD with the noisy a rather than a in the 11th commend of the MatLab?