Underfitting: high bias, a model too simple for the signal, so training and validation error are both high and close to each other. Add capacity or better features, weaken the regularization, or train longer. More data will not rescue a model that cannot fit what it already has.
Card 1 of 259. Answer: Underfitting: high bias, a model too simple for the signal, so training and validation error are both high and close to each other. Add capacity or better features, weaken the regularization, or train longer. More data will not rescue a model that cannot fit what it already has.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
Underfitting. A model underfits when it is too simple or too constrained to capture the real pattern in the data, so it performs poorly on the training set itself and on new data too. In bias-variance terms it has high bias, and more training data will not fix it.
Example: fit a straight line to y = x² with x spread evenly from −3 to 3. Because the curve is symmetric, the best line is flat at about y = 3, its training R² is close to 0, and validation error is about as high. Learning curves show the signature: training and validation error converge quickly and plateau at a high level with little gap.
Fixes run the opposite way from overfitting: add features or interactions, use a more flexible model, weaken regularization or train longer. Interviewers often ask how you tell underfitting from noisy data. Compare training error with a baseline or human-level error: if it is far above what is achievable, the model is underfitting.
