Table 4.

Explanation Consistency (C) and JSD Explanation Consistency ($C_{\mathit{JSD}}$) for the baseline models and explanation ensembles across all tasks tested.

Dataset (Task)	Base mModel $\mathbf{C}$	Explanation ensemble $\mathbf{C}$	Base model ${\mathbf{C}}_{\mathbf{JSD}}$	Explanation ensemble ${\mathbf{C}}_{\mathbf{JSD}}$
BCW	0.12282	0.4450 (262%)	0.24682	0.273065 (11%)
Diabetes (Classification)	0.58550	0.93697 (60%)	0.51667	0.543646 (5%)
Diabetes (Regression)	0.52691	0.600067 (13%)	0.35389	0.65568 (85%)
Codon Usage (DNA)	0.34279	0.5564 (62%)	0.28114	0.340558 (21%)
Codon usage (Kingdom)	0.22220	0.29387 (32%)	0.34702	0.39391 (14%)
MIMIC-IV	0.02433	0.10111 (315%)	0.15518	0.17912 (15%)

The percentage increase from baseline C ($C_{\mathit{JSD}}$) to explanation ensemble C is shown in brackets.

Significant values are in [bold].