SSE vs. SEE

[MJLU280]

Can someone please explain the difference between SSE and SEE as they both seem to be measuring the unexplained variation between predicted Y and actual Y in a regression equation. Thanks.

 

[Alladin]

http://www.youtube.com/watch?v=dJR1WqeBgCg

 

[Thecodont]

I have not reviwed it since i read it but as i recall, SSE is the sum of squared errors while SEE is the Sq Root of MSE. The point is that like R^2 it is a measure of fit since its the st dev of the unexplained errors. As a result, unlike R^2 you want a low SEE since that indicates a small amount of unexplained error

 

[daneshghara]

SEE=square(SSE/n-k-1))

 

[Thecodont]

daneshghara wrote:
SEE=square(SSE/n-k-1))

you mean to say the square root rather than the square SSE is already squared units. Thats why its called sum of squared errors

 

[MJLU280]

Thank you Alladin! This was driving me insane, great explanation by David Harper. I will check out some of his other tutorials.
- Mike

 

[Alladin]

Hey Mike,glad you found it useful, the bionicturtule on youtube is a great resource in my opinion!

 

[iteracom]

One day you’ll all look back and say, oh yeah I studied that but I don’t remember what it was

 

[Alladin]

iteracom wrote:
One day you’ll all look back and say, oh yeah I studied that but I don’t remember what it was

5 mins after the exam i forget what it was all about