Table 1. The jackknife success rates achieved by iATC-mHyb and ATC-mISF on the benchmark dataset $\mathbb{S}$ of Eq.1 (cf. Supporting Information S1).

Predictor	Five metrics for multi-label systema
	Aiming↑b	Coverage↑b	Accuracy↑b	Absolute true↑b	Absolute false↓c
iATC-mISFd	67.83%	67.10%	66.41%	60.98%	5.85%
iATC-mHybe	71.91%	71.46%	71.32%	66.75%	2.43%

^aSee Eq.12 for the definitions of the five metrics used to measure the prediction quality for multi-label systems [3].

^bThe upper arrow means that the larger the rate the better the prediction quality is.

^cThe down arrow means that the smaller the rate the better the prediction quality is.

^dThe predictor proposed in [4].

^eThe predictor proposed in the current paper.