... between X and rMSE Focusing for now on the circles (pinpointing the ZM, fZM and GIGP models), we see that there is instead a negative correlation between goodness of fit on the training set and quality ... overfitting the training set, leading to poorer performance than the simpler ZM on unseen data We turn now to what we think is the main cause for this overfitting Non-randomness and echoes The results ... on the training set and quality of prediction, as illustrated for V prediction at 3N0 by the triangles in Figure (again, the patterns in this 10 In looking at the V1 data, it must be kept in mind,...