Skip to main content
. 2021 Mar 22;37(18):2889–2895. doi: 10.1093/bioinformatics/btab192

Fig. 5.

Fig. 5.

(a) Performance in deconvolution accuracy (Spearman) and (b) detection accuracy (AUC) of an ensemble based on the median prediction of all possible combinations of a given size of the top nine algorithms plus the benchmark. The median performance of the ensemble tends to increase with its size. However, the maximum performance in both metrics tends to plateau (or even decrease) after the ensemble reaches a size equal to three