Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

How, if at all, can one compare the "fit" of a simple linear vs. non-linear regression model to observed data?

I apologize if I didn't search long/hard enough for the answer, but I cannot find anything concrete.


share|improve this question
What software are you using? – Michelle Jan 26 '12 at 23:39
Can you share your purpose in comparing the fit of the different models? To develop a predictive model? To describe a particular data set? To provide evidence for or against a particular theory? You might choose different comparison approaches depending on your purpose. – Anne Z. Jan 27 '12 at 3:05
Anne, at this point it is simply for description. I would like to quantitatively be able to say that a non-linear fit is preferred. – Patrick Jan 28 '12 at 17:21
Michelle, I use R. – Patrick Feb 3 '12 at 17:20
Would using AIC/BIC or the LogLik be sound statistically to make a choice between non-linear and linear models on the same dataset? – user9974 Mar 19 '12 at 23:33

Perform cross-validation on each model on a development set to find the best hyper-parameters / accuracy estimate for each. Then check that the accuracy estimates are still valid on some held out data that you did not use during the model selection / parameter tuning phase.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.