figshare
Browse

sorry, we can't preview this file

pcbi.1008878.s003.eps (88.66 kB)

AKLIMATE versus alternative ways of ensembling/stacking component RFs.

Download (88.66 kB)
figure
posted on 2021-04-16, 17:42 authored by Vladislav Uzunangelov, Christopher K. Wong, Joshua M. Stuart

To ensure fair comparison, the component RF set for each prediction task was restricted to AKLIMATE’s best RFs Δ*. Ensemble superlearner component RFs—learning a regularized linear regression on predictions from component RFs; ensemble average component RFs—taking the average of the predictions of component RFs; top component RF—only using predictions from the top ranked RF. (A) UCEC TCGA MSI prediction. (B) METABRIC Breast Cancer survival prediction. (C) Achilles shRNA knockdown prediction. P-values for paired Wilcoxon signed rank test.

(EPS)

History