Development and validation of a prediction model estimating the 10-year risk for type 2 diabetes in China
Fig 2
Feature selection using the least absolute shrinkage and selection operator (LASSO) binary logistic regression model.
(A) Turning parameter (λ) selection in the LASSO model used 10-fold cross-validation via minimum criteria. The AUC curve was plotted versus log (λ). Dotted vertical lines were generated at the optimal values using the minimum criteria (the 1-SE criteria). A λ value of 0.0003 with log (λ), -8.099 was chosen (1-SE criteria). (B) LASSO coefficient profiles of the 8 features. A coefficient plot was generated against the log (λ) sequence. (C) Features that were selected for the second model. Similarly, dotted vertical lines were drawn at optimal values and λ value of 0.001 with log (λ) -6.520 was chosen. (D) LASSO coefficient profiles of the 17 features are shown. A coefficient plot was generated against the log (λ) sequence. (E) A list of features selected for the third model. A λ value of 0.002 with a log (λ) of -6.227 was chosen. (F) A list of LASSO coefficient profiles of the 20 features. (G) A list of features selected for the fourth model. A λ value of 0.001 with a log (λ) of -6.526 was chosen. (H) LASSO coefficient profiles of the 5 features are listed.