I'm a legally blind student who does not see the board in the regular classroom. I'd like to thank you for sharing these nice videos. I'd recommend them to all of my classmates!
Great explanation. You forgot to remind that X'X is indeed a symmetric matrix, to make the analogy with A matrix that was symmetric in previous video. That's from going to last line from the second last. Cheers!
Hello, I have a question about x'Ax matrix the derivative is equal 2x'x this applies only when our matrix is symmetrical, but in example if we have 4 features and 1k tests how this matrix is symmetric is not square, I am bit confused this in real life example in ML
X might not be symmetric but X^T X ensures it is symmetric. Just build your design matrix X as n x p sized with n beeing the number of observations and p = k + 1 with k beeing the number of regressors.
Is there a simpler way to remember the rules when differentiating because it will be long to do all this in an exam just to differentiate, so is there like a simple rule to follow? Thank you for your videos!
Hi , k have a doubt. i found in theory being w= y'Ax, so dw/dy= x'A' , hence in this video being the third term S= -B'x'y , so the dS/dB should have been -y'x? sorry if my question is too basic.
In this video, you say that dS/dB' for the -B'X'y term is just -X'y because dy/dX' = a. But didn't we find in the previous video that dy/dX' = a'? In that case, shouldn't the derivative be -y'XB? I realize that in the end, it doesn't make a difference, but I'm just trying to clear up my understanding...
Hi, thanks for your message. I think say in the video that the derivative of y wrt x is a - which makes sense dimensionally. Hence, hopefully everything else should follow. Hope that helps. Best, Ben
Hi, thanks a lot for your videos! They are helping a lot! :) But could someone tell me why he is underlining some of the variables? Is there a meaning of underlining (straight line or this waved line)?
![I.N "HALLUCINATION" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/n5B5q1Hwt_U/mqdefault.jpg)
I'm a legally blind student who does not see the board in the regular classroom. I'd like to thank you for sharing these nice videos. I'd recommend them to all of my classmates!
Hi, many thanks for your message. Glad to hear it helped! Best, Ben
God Bless You, Ben Lambert
Great explanation. You forgot to remind that X'X is indeed a symmetric matrix, to make the analogy with A matrix that was symmetric in previous video. That's from going to last line from the second last. Cheers!
Puff, you are awesome, thank you very much!
Great material put together!
indeed, really helpful contents! thx from Korea
Thank you for explaining in such detail
Hello,
I have a question about x'Ax matrix the derivative is equal 2x'x this applies only when our matrix is symmetrical, but in example if we have 4 features and 1k tests how this matrix is symmetric is not square, I am bit confused this in real life example in ML
X might not be symmetric but X^T X ensures it is symmetric. Just build your design matrix X as n x p sized with n beeing the number of observations and p = k + 1 with k beeing the number of regressors.
Outstanding work you helped me a lot!
Isn't Beta-Hat=[Beta-Hat_0, Beta-Hat_1, Beta-Hat_2, ... , Beta_Hat_p]', so aren't the dimensions 1 x (p+1) ?
thanks these are sooo helpful!!! but why is this considered graduate when im doing this in an undergraduate course..
Is there a simpler way to remember the rules when differentiating because it will be long to do all this in an exam just to differentiate, so is there like a simple rule to follow?
Thank you for your videos!
brilliant !!! This is very helpful !!
Very helpful.
Really helpful! Thanks a bunch!
You da bomb Ben!
Hi Ben, I would love to see the analagous derivation for multiple correlated output variables--for multiple-output regression
its very helpfill sir...tnx for uploading
So for the last one - Is the derivative of B'X'X the same as the derivative of X'XB?
Hi , k have a doubt. i found in theory being w= y'Ax, so dw/dy= x'A' , hence in this video being the third term S= -B'x'y , so the dS/dB should have been -y'x? sorry if my question is too basic.
Thank you soo much for your video
Can we use the same matrix form when we have categorical variables?
In this video, you say that dS/dB' for the -B'X'y term is just -X'y because dy/dX' = a. But didn't we find in the previous video that dy/dX' = a'? In that case, shouldn't the derivative be -y'XB? I realize that in the end, it doesn't make a difference, but I'm just trying to clear up my understanding...
Hi, thanks for your message. I think say in the video that the derivative of y wrt x is a - which makes sense dimensionally. Hence, hopefully everything else should follow. Hope that helps. Best, Ben
Hi, thanks a lot for your videos! They are helping a lot! :)
But could someone tell me why he is underlining some of the variables? Is there a meaning of underlining (straight line or this waved line)?
Hi, thanks for your comment. The variables with straight lines below them are vectors. Those with curvy lines are matrices. Best, Ben
Saving my life
Thanks a lot!!
is beta 0 (the intercept) included in the beta hat vector?
beta 0 is included in the vector of beta estimates.
Thank you so much but i still could not understand why the differentiation of X'a=a
4:25
ruclips.net/video/iWxY7VdcSH8/видео.html
@@ritishmaram3326 THANK YOU SOOOOOO MUCH!!!