Underfitting
Definition
When a model is too simple or trained too little to capture the pattern in the data, so it does poorly even on its training data.
In plain English
Underfitting is the opposite of overfitting. The model hasn’t learned enough, so it makes rough, general guesses that are wrong on both the data it practiced on and data it hasn’t seen.
Common fixes are using a more capable model, giving it more useful features, or training it for longer.
Example
Drawing a single straight line through house prices that clearly rise faster for bigger houses. The line misses the curve everywhere, on old and new data alike.
Draft, not yet reviewed by a person · Updated Nov 2026