What is underfitting? How can you address it?

Question

Accepted Answer

Underfitting happens when a model is too simple to capture the underlying pattern in the data. It performs poorly on both training and validation data. How to detect — both training and validation errors are high; the model never reaches a satisfactory loss even on training data. Causes — model with too little capacity, missing relevant features, excessive regularization, too few training epochs. How to fix: Use a more complex model (deeper tree, more layers). Add more relevant features or interaction terms. Reduce regularization . Train longer (more epochs). Reduce noise in the data. Underfitting is the mirror of overfitting. The fix is roughly the opposite — more capacity, fewer constraints. Watch for the giveaway signal: training error itself is high (the model couldn't even fit what it was shown).