Skip to main content

Table 4 UCI ML madelon dataset validation

From: binomialRF: interpretable combinatoric efficiency of random forests to identify biomarker interactions

Model Model size Run time Precision Recall
VarSelRF 23 (13) 129 (21) 0.56 (0.01) 0.56 (0.02)
VSURF 3.5 (1.4) 321 (267) 0.56 (0.02) 0.56 (0.03)
binomialRF 17.1 (3.9) 5.6 (2.2) 0.55 (0.02) 0.55 (0.01)
Vita 13 (5.68) 1007 (1220) 0.55 (0.02) 0.55 (0.02)
Boruta 2 (2) 139 (45) 0.54 (0.03) 0.56 (0.04)
Perm 240 (13) 269. (329) 0.56 (0.08) 0.54 (0.01)
AUCRF 31 (30) 33 (7.5) 0.55 (0.04) 0.54 (0.02)
RFE 81 (4.2) 20 (1.4) 0.54 (0.06) 0.54 (0.01)
EFS 20 (8.3) 2617 (2126) 0.53 (0.02) 0.54 (0.02)
PIMP 1.7 (1.3) 482 (128) 0.50 (0.04) 0.50 (0.01)
  1. The algorithms in Table 1 were tested and compared using the Madelon benchmark dataset from UCI (described in Methods). Mean (standard deviation) results are shown and ranked according to decreasing harmonic mean of precision and recall of variables. Top accuracies are bolded