Hi,
When we have a model and we increase it’s flexibility - Isn’t it equally likely that the model will still under-fit (but slightly less) or is it more likely that it will over-fit ?
You have to consider somehow a range of low-flexbility to high-flexbility. One would expects that a model will underfit for low-flexibility and as it becomes more flexible, it might rich a generalization range before falling into an overfitting regime.
It could happen that some models will never be flexible enough to reach the generalization and will remain in the underfitting regime.
However, we present the former general idea but it is good to have in mind the second one as well.
1 Like
Thank you for the prompt reply 