Figure 2.

Diagrams illustrate under- and overfitting. Underfitting occurs when the fit is too simple to explain the variance in the data and does not capture the pattern. An appropriate fit captures the pattern but is not too inflexible or flexible to fit data. Overfitting occurs when the fit is too good to be true and there is possibly fitting to the noise in the data. The axes are generically labeled feature 1 and feature 2 to reflect the first two elements of the feature vector.