Reply
Fri 26 Feb, 2010 07:32 am
In regression analysis the coefficient of determination (R^2) is good (0.8). That means 80% correlation exists between dependent & independent variables. But, acual Y values & predicted Y values (dependent variable is Y that is to be estimated) many values of predicted Y are not in the range of 80% of that of actual Y values i.e. many predicted Y dont lie in between limits 0.8*Yactual to 1.2*Yactual.
So, for good results, what we should look for, how many predicted Y's (in percentage)should lie in what limits (like 80% as stated above)?
I'm really confused. So, please help me.
@asp2801,
You don't want R^2, you want the individual confidence interval. You should be able to get the standard error for the correlation and if the residuals are normally distributed, you can use that to calculate a confidence interval.
@engineer,
Thank you very much for your help! please tell me hoe to use standard error to find confidence interval?
Stumble It!
Tweet This
Bookmark on Delicious
Share on Facebook
Share on MySpace