5.7 Advantages of k-fold Cross-Validation over LOOCV

  • The main advantage of k-fold over LOOCV is computational.
  • However, there are other advantages related to the bias-variance tradeoff.
  • The figure below shows the true test error for the simulated data sets from Chapter 2 compared to the LOOCV error and k-fold cross-validation error.
The estimated test errors for LOOCV and k-fold cross validation is compared to the true test error for the three simulated data sets from Chapter 2. True test error is shown in blue, LOOCV error is the dashed black line, and 10-fold error is shown in orange.

Figure 5.4: The estimated test errors for LOOCV and k-fold cross validation is compared to the true test error for the three simulated data sets from Chapter 2. True test error is shown in blue, LOOCV error is the dashed black line, and 10-fold error is shown in orange.