I have gone through the model selection process for my linear model prior and post model transformation. I would like to know whether it is ok for me to compare the AIC values for both models given the AIC value prior transformation is approximately 180 where the AIC value post transformation is approximately -140. Do I definitively compare and take the lowest (negative) value or do I take the lowest absolute AIC value?
Asked
Active
Viewed 38 times
2 Answers
2
you can't compare AIC values of models with different left hand side, e.g. transformed and untransformed variables.

Aksakal
- 55,939
- 5
- 90
- 176
0
From this stackexchange answer,
For model comparison, the model with the lowest AIC score is preferred. The absolute values of the AIC scores do not matter.
However, in your case, the lowest AIC value is the same as the lowest absolute AIC value. So you should definitely pick the model that gives you -140 AIC.

Josh Bone
- 123
- 7