“An optimal model is a ‘Goldilocks’ model. It is large enough that it can reliably detect potentially complex predictive relationships in the data, but not so flexible that it is dominated by overfit and suffers out-of-sample.”
- S.B. Gu discussing machine learning and econometrics
A great way to think about a model is to use the Goldilocks analogy. A model should not be too complex or too simple. Too complex with too many variables and we have an overfit problem. Too few variables and we are left with a problem that an unspecified factor will drive predictions. While many would like to believe that building models is a science, there is a lot art to finding the right model. A model will have the personality of the author.
No comments:
Post a Comment