Skip to main content

Table 5 Features remaining after both feature-selection stages for models in the Stage 2 analysis

From: A comparison of feature selection methodologies and learning algorithms in the development of a DNA methylation-based telomere length estimator

Model

Features remaining after initial feature-selection stage

Features remaining after elastic net stage

Baseline (elastic net)

–

832

F-test (FDR: 0.01)/elastic net

14,453

3893

F-test (FDR: 0.05)/elastic net

34,365

5448

Gradient boosting/elastic net

614

446

Pearson correlation/elastic net

750

251

Mutual information/elastic net

6500

407

Linear SVR/elastic net

8500

4945

Random forest Regression/elastic net

3250

1059

PCA/elastic net

1349*

111*

  1. Note that there is no initial feature-selection stage for the baseline model. *for PCA with elastic net, the number of principal components is shown, as PCA does not return a set of explicit features