Question

Question #8f072

Answer 1

Depends on the model, generally the formula is:

`epsilon=y-E[y]`

Where `E[y]` is the value of `y` predicted by the model from the observed values of `x`.