How to calculate sum of squared errors when your regression has no constant

PTRUSCOTT
PTRUSCOTT used Ask the Experts™
on
Dear Experts,
I have an application that requires me to work out the R-Squared statistic of a regression from scratch.  I got the formula from the following Web page:

http://en.wikipedia.org/wiki/Coefficient_of_determination

This works fine for a normal regression model with a constant.  If you select the noconstant option in either Excel or Stata then the Sum of Squared differences is totally different from the result of the formula shown on the Wikipedia page above.

Can anyone show me a link to the correct formula where there is no intercept?
Best Wishes,
Phil
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
With a straight line there is always an intercept unless the straight line is parallel to one of the axes.
ozo
Most Valuable Expert 2014
Top Expert 2015

Commented:
By "no constant" do you mean constant = 0 ?

Author

Commented:
Yes. I mean constant = zero. "Noconstant" is the keyword you use in Stata

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial