[R] Predicting confidence intervals for fitted values for non-linear regression

Mills, Kathryn (NIH/NIMH) [F] kathryn.mills at nih.gov
Tue Mar 20 11:10:31 CET 2012


I am interested in calculating the confidence intervals for fitted values for non-linear regressions. For instance, I have used the nlme package to generate my non-linear model.

summary(lme(myvariable~age+age.sq+age.cu, data=my.matrix, random=~1|Name))

I would like to use the model generated from the data in my.matrix to predict the confidence intervals for new ages. Someone posted a question similar to mine a few years ago, but the website someone sent in response to the question is no longer in use. I will paste their question below because I think it is clearer than mine.

Thank you for your help,

"I am interested to calculate confidence interval for fitted values in general for non-linear regressions. Lets say we have y=f(x1,x2,..xN) where f() is a non-linear regression. I would like to calculate a confidence interval for new prediction f(a1,..,aN). I am aware of techniques for calculating confidence intervals for coeffiecients in specific non-linear regressions and with them then to calculate confidence interval for the predicted value. ...

Any references to the literature or R packages would be very welcome."

More information about the R-help mailing list