In the following R code where the dataset Fin has the variables SPY, X1, X2, X3 which of the following commands would generate the proper mean squared error for the test sample:

attach(Fin) SubData=data.frame(SPY, X1, X2, X3) len=dim(Fin)[1] train=sample(len,*.7,replace=FALSE) full=lm(Y~.,data=SubData)

A) mean((Fin$SPY-predict(full,Fin))[-train]^2)

B) mean((Fin$SPY-predict(full,Fin))^2)

C) None of the other options

D) mean((Fin$SPY-predict(full,Fin))[train]^2)