Table 2 Leave-One-Out Cross-Validation (LOOCV) of robust regression models

From: Motor learning induces myelin-related white matter changes revealed by MRI-based in vivo histology

model

no. of predictors

R2 (original model)

LOOCV MSE

LOOCV RMSE

1 - (SSE_model ÷ SSE_null)

Version 1

1

.19

22.36

4.73

0.07

Version 2

5

.36

26.46

5.14

-0.10

  1. Model accuracy was quantified using mean squared error (MSE), root mean squared error (RMSE), and the ratio of the model’s sum of squared errors (SSE_model) to a null model always predicting the training mean (SSE_null). Positive ratios indicate predictive power; negative ratios indicate overfitting.
Back to article page