GoCareerGoCareer prep studio
LoginSign up
← All tutorials

GoCareerGo Tutorials

Machine Learning Tutorial

Supervised learning, classification, neural networks and core ML algorithms.

Overfitting vs Underfitting

Overfitting memorizes the training data (poor generalization); underfitting fails to capture the pattern at all (poor everywhere).

  • Overfitting — great training accuracy, poor test accuracy; the model is too complex relative to the data
  • Underfitting — poor accuracy on both train and test; the model is too simple to capture the pattern
  • Fixes for overfitting — more data, regularization, simpler model, dropout, early stopping
  • Fixes for underfitting — more features, more complex model, less regularization
Tip

Draw the classic 'training vs validation loss curve diverging' picture in your head — it's the mental model interviewers expect.