Overfitting and Underfitting
Question: What are the differences between overfitting and underfitting?
- Overfitting comes from too little data.
- Underfitting happens if your model is too simple for the data.
Example: When fitting a model, you need to make reliable predictions. Overfitting causes those predictions to be too sensitive to noise in the data, while underfitting simplifies the data too much– possibly from too few dimensions. We counter those problems by creating a training and test section of the data, possibly with cross-validation. Training the model on training and then using the test section of the data will help us see how much overfitting occurs.