Whenever you're ready, walk me through how you'd approach this.
You are comparing a few supervised learning models and trying to understand why some underfit while others overfit.
What is the bias-variance trade-off?