Fitted values

A linear model provides a prediction of y for any values of x and z. For the i'th individual, the prediction is called its fitted value,

fitted value,    =  b0 + b1 xi + b2 zi

Residuals

The difference between the fitted value and the actual response is called the individual's residual.

residual,   ei  =  yi

Residuals correspond to vertical distances between crosses in a 3-dimensional scatterplot and the plane representing the model.