Skip to main content

Table 3 Mean of the number of informative features (% of uninformative features) selected in final model over 300 simulations in the training sets

From: Combined performance of screening and variable selection methods in ultra-high dimensional data in predicting time-to-event outcomes

Selection approach Weak signal Strong signal
n = 150 n = 300 n = 150 n = 300
ISIS-LASSO 0.373 (64.1%) 5.913 (1.6%) 1.920 (35.3%) 6.000 (1.0%)
ISIS-ALASSO 0.383 (59.9%) 5.897 (0.9%) 1.983 (30.0%) 6.000 (0.5%)
ISIS-RSF 0.343 (88.3%) 5.817 (3.3%) 1.657 (61.1%) 5.973 (0.9%)
SIS 1.973 (89.4%) 0.160 (5.5%) 1.720 (70.7%) 0.023 (0.4%)
SIS-LASSO 0.303 (56.4%) 2.960 (2.8%) 0.757 (33.4%) 4.357 (0.4%)
SIS-ALASSO 0.303 (57.3%) 2.960 (1.6%) 0.757 (29.6%) 4.357 (0.3%)
SIS-RSF 0.293 (88.8%) 2.880 (5.3%) 0.683 (69.4%) 4.317 (0.2%)
PSIS 4.577 (96.5%) 5.943 (95.0%) 5.437 (95.8%) 5.987 (94.9%)
PSIS-LASSO 4.547 (93.4%) 5.943 (92.4%) 5.437 (91.3%) 5.987 (90.8%)
PSIS-ALASSO 4.180 (91.9%) 5.943 (87.5%) 5.403 (86.5%) 5.987 (78.7%)
PSIS-RSF 2.183 (94.8%) 3.717 (73.8%) 3.703 (90.7%) 4.973 (63.3%)
LASSO 4.671 (80.0%) 6.000 (89.7%) 5.990 (89.4%) 6.000 (94.4%)
ALASSO 5.731 (65.0%) 6.000 (86.2%) 6.000 (81.1%) 6.000 (90.9%)