[R] Is there any criteria to compare different models?

[Saman Zahid]

Hi,

I am trying to make comparison between the lm, gam(mgcv) and lme(nlme)
models. The problem is I am not able to find a single criterion on which
these 3 models are comparable.

I am using method "ML" for both gam and lme, yet the AIC and BIC formulas
differ due to the degrees of freedom.

Is there any statistical way to compare these three models?

I will really appreciate your help.

Thanks
Saman Zahid

	[[alternative HTML version deleted]]