Fig. 2.

Analysis of LASSO regression models for selecting factors. (A) LASSO coefficients for 26 candidate variables. (B) The optimal penalty coefficient ($\lambda$) was found using cross-validation and the minimum criterion. The cross-validated error within one standard error (1-s.e.) of the minimal error is shown on the right, while the minimum error is shown on the left. Abbreviations: LASSO, Least absolute shrinkage and selection operator.