Fig. 3
From: Predicting the recurrence of noncoding regulatory mutations in cancer
Proportion of positive votes by 1000 decision trees of the random forest. a-b Proportion of votes by the classifier trained with the raw recurrence model (≥4) for mutations in the test samples. c-d Proportion of votes by the classifier trained with the significant recurrence model (p-value < 5 × 10−6) for mutations in the test sample. a, c Voting for mutations having no (0), moderate (2 or 3) recurrence, or high (≥4) raw recurrence. b, d Voting for mutations whose recurrence significance was low (p ≥ 0.01), moderate (5 × 10−6 ≤ p < 0.01), or high (p < 5 × 10−6)