Automatic design of decision-tree induction algorithms tailored to flexible-receptor docking data

BMC Bioinformatics

Table 2 Results of Experiment 1

Measure	Ligand	HEAD	C4.5	$\hat{μ}$	${\hat{σ}}^{2}$	t
Accuracy	ETH	0.71 ± 0.02	0.62 ± 0.02	0.08	0.0008	6.25
	PIF	0.86 ± 0.00	0.80 ± 0.02	0.06	0.0004	6.22
	TCL	0.65 ± 0.02	0.57 ± 0.02	0.08	0.0003	9.12
	INH	0.84 ± 0.01	0.79 ± 0.01	0.05	0.0001	8.28
	JPM	0.72 ± 0.02	0.65 ± 0.02	0.07	0.0004	7.62
Tree Size	ETH	45.40 ± 9.88	539.80 ± 29.64	494.40	1174.04	31.48
	PIF	6.80 ± 0.63	283.00 ± 39.97	276.20	1603.95	15.00
	TCL	38.20 ± 8.17	588.00 ± 37.39	549.80	1549.73	30.40
	INH	12.20 ± 3.91	278.60 ± 24.80	266.40	568.71	24.31
	JPM	28.20 ± 6.48	483.80 ± 19.76	455.60	408.71	49.05

Table2 presents the classification accuracy and tree size of HEAD-DT and C4.5 in the ETH, PIF, TCL, INH, and JPM data sets. HEAD-DT results were obtained by running a single decision-tree algorithm tailored to the NADH data set.

ISSN: 1471-2105