| Author |
 Topic  |
|
|
erickzhou
17 Posts |
 Posted - 07/18/2010 : 11:34:23 PM
|
Origin Ver 8. and Service Release (Select Help-->About Origin):
Operating System: WinXP
First question:
I have fitted my data by 2-D polynomial surface, and the "Adj. R-Square" is 0.92173. However, I don't know well about the equation to get adjusted R-square error. The equation is R^2=1-(Rss/dferror)/(Tss/dftotal), I don't know the meaning of dferror and dftotal. Can you give me some explationations and references. The detailed information of this equation can be found in the chapter of "goodness of fit" (http://www.originlab.com/www/helponline/Origin/en/UserGuide/Goodness_of_Fit.html).
Second question:
In the chapter of "Graphic Residual Analysis"
(http://www.originlab.com/www/helponline/Origin/en/UserGuide/Graphic_Residual_Analysis.html), the residual vs. independent plot was given for the case of curve fitting. However, no quantitive results was given about residual analysis, they only told me to judge from the plots.
Furthermore, in the chapter of "Interpreting results" of the reference of "Marko Ledvij. Curve Fitting Made Easy."(The Industrial Physicist. Apr./May 2003. 9:24-27.), residual analysis was not used to expalin how well the model fits the data.
From above information, can I say that if I have already known the true model of my data, the "Adj. R-Square error" is enough for assessing the model.
|
|
|
larry_lan
China
Posts |
Posted - 07/19/2010 : 9:26:24 PM
|
You can found dferror and dftotal here, and the value can found form ANOVA Table in the report worksheet.
This book can be help to learn more about regression.
Thanks,
Larry |
 |
|
| |
Topic |
|
|
|
Questions on Goodness of fit and Graphic Residual