Underfitting

Underfitting is the opposite failure: the model is too simple or too constrained to capture the real structure in the data, so it does poorly even on the training set, let alone new data. Fitting a straight line to data that clearly curves is the textbook case.

Where overfitting is too much flexibility, underfitting is too little — a high-bias model that misses the signal. The fix is to give the model more capacity: add relevant features, include interactions or curved (polynomial) terms, or choose a more expressive model. The art of modeling is steering between this and overfitting.