- When you spend too much time on the problem, especially one where the dataset is small, and CV is hard to make, you could be trying too hard
- this means you’ll make features that improve CV, but overall, doesn’t improve the generalization of your model
- TLDR; spending too much time ⇒ making overfitted models