Before we get into modelling and try to figure out what the trade-off is, let's understand what bias and variance are from the following diagram:
There are two types of errors that are developed in the bias-variance trade off, as follows:
- Training error: This is a measure of deviation of the fitted value from the actual value while predicting the output by using the training inputs. This error depends majorly on the model's complexity. As the model's complexity increases, the error appears to plummet.
- Development error: This is a measure of deviation of the predicted value, and is used by the development set as input ...
