In Table 10.2 on page 433, we see that the ordering of the three methods with respect to mean absolute error is different from the ordering with respect to test set R2. How can this be?

10.6 When to Use Deep Learning 433 Model # Parameters Mean Abs. Error Test Set R2 Linear Regression 20 254.7 0.56 Lasso 12 252.3 0.51 Neural Network 1409 257.4 0.54 TABLE 10.2. Prediction results on the Hitters test data for linear models fit by ordinary least squares and lasso, compared to a neural network fit by stochastic gradient descent with dropout regularization. Coefficient Std. error t-statistic p-value Intercept -226.67 86.26 2.63 0.0103 Hits 3.06 1.02 3.00 0.0036 Walks 0.181 2.04 0.09 0.9294 CRuns 0.859 0.12 7.09