In
statistics the
mean squared prediction error of a
smoothing or
curve fitting procedure is the expected value of the squared difference between the fitted values
![](http://info.babylon.com/onlinebox.cgi?rt=GetFile&uri=!!ARV6FUJ2JP&type=0&index=3575)
and the (unobservable) function
g. If the smoothing procedure has
operator matrix L, then