What are the criteria to select the best growth model ??
(Adjusted R2, AIC, or if any...)
Whether AIC can be calculated for non-linear regression growth model
In order to assess goodness of fit for a non-linear model, you can:
1) plot raw data against fitted one and compare
2) use S measure
3) AIC
4) BIC
R^2 is suitable for linear model.
AIC, BIC, and some pseudo r-square measures are based on the log likelihood and the number of terms and observations in the model. If you are able to extract the log likelihood from a model, these statistics can be calculated.
Sometimes AIC etc. are expressed in terms of deviance instead of log likelihood.
Thanks Dr. Faye Anderson and Dr. Salvatore S. Mangiafico for necessary information.
If the model is nested model you log likelihood is preferable unless AIC and BIC are preferable criteria. Caution: Don't use log likelihood to compare the models if the model is not nested models.
- Badges
- Science topic
- Similar topics
- Applied Mathematics
- Calculations