# How to compute goodness of fit for a linear model in R

I have fit a linear model using the lm function in R...

model <- lm(trans.baseline.CD4 ~ hiv\$Julian.Date)


... and I would like to assess the quality of the model's fit. Is there a function in R that will do this? Alternatively, I found a formula for goodness-of-fit involving the sum of squared residuals given the null and alternative hypotheses, but I don't know how to get these values either. Any pointers?

Thanks!

-

It all starts with

 summary(model)


after your fit. There are numerous commands to assess the fit, test commands, compare alternative models, ... in base R as well as in add-on packages on CRAN.

But you may want to do some reading, for example with Dalgaard's book or another introduction to statistics with R.

-
(+1) Linear Models with R, by Julian J. Faraway is also a good starting point. – chl Nov 13 '10 at 18:07
Yeah, I guess there's really no escaping me actually digging down and figuring out what exactly I want to compute. I was just hoping there was some sort of standard first approach to use. – Daniel Standage Nov 13 '10 at 21:14
Yup. With great freedom comes great responsibility. It is easy to just mechanically compute something. It may be much harder to come up something meaningful. Damn No Free Lunch theorem again... – Dirk Eddelbuettel Nov 13 '10 at 22:15