Underfitting
A phenomenon in machine learning in which a model is too simple to learn the basic patterns of the training data, manifesting as high error on both training and test data. It typically requires increasing model complexity or adding features.
Timeline (last 90 days)
Underfitting means the model is too simple to learn even the basic patterns of the training data, with both training error and test error high.
A decision boundary that is too simple may lead to underfitting, while one that is too convoluted may lead to overfitting.
Methods for addressing underfitting include adding features, switching to a stronger model, and performing feature engineering; methods for addressing overfitting include adding regularization, reducing features, increasing data, or using cross-validation.
All Facts (3)
Underfitting means the model is too simple to learn even the basic patterns of the training data, with both training error and test error high.
50%UnverifiedA decision boundary that is too simple may lead to underfitting, while one that is too convoluted may lead to overfitting.
50%UnverifiedMethods for addressing underfitting include adding features, switching to a stronger model, and performing feature engineering; methods for addressing overfitting include adding regularization, reducing features, increasing data, or using cross-validation.
50%