Hi All,

Assume I have two variable set as: 'Actuals' and their "Expected" values. My
problem is to compare how good my fit is provided I dont have any prior
infirmation about the type of fitted model?

I also dont want to assume anything about the actuals/expected data. The
method which till now I was using was 'sequential graphs and pivot
charting'. But now I want some more lucid statistical measures for the same.
I hope relative error, % error, sum of sq due to error, R square can be
applied without assuming anything about the model and the concerned
variable.

Please correct me if I am wrong and pray suggest some more method for
goodness of fit.

Note: chi sq cannt be applied as it assumes categorical data. paired t-test
assumes normality so is ruled out from this picture.

Thanks in advance

Huma Zaidi
91-9886620871

_________________________________________________________________
Fall in Love... Get married! Join FREE!
http://www.shaadi.com/ptnr.php?ptnr=msnhottag