Although there wasn’t a clear trend for base-learner pair-wise diversity with time-group across base-learners (see Cohen k-statistic in Supplementary Figs. 35 , 36 ), the JTG2L BMA stack did, in fact, outperform the best base-learner in the training set (Fig. 1a ), and any of the remaining stacks (see Supplementary Figs. 27 , 28 ).
← all excerpts
Serum biomarker-based early detection of pancreatic ductal adenocarcinomas with ensemble learning.
1
—
—