Table 5.

Performance comparison of our method with different classifiers and different essential and non-essential gene rates on InWeb_IM dataset.

Methods	Precision	Recall	SP	NPV	F-measure	MCC	Accuracy	AUC	AP
DNN (1 layer, 1:6)	0.355	0.772	0.206	0.877	0.261	0.103	0.712	0.499	0.206
DNN (3 layers, 1:6)	0.350	0.255	0.921	0.881	0.295	0.202	0.826	0.483	0.221
DT (1:6)	0.584	0.588	0.930	0.931	0.586	0.517	0.881	0.759	0.616
NB (1:6)	0.456	0.697	0.861	0.945	0.551	0.473	0.837	0.877	0.615
KNN (1:6)	0.778	0.564	0.973	0.930	0.654	0.617	0.914	0.888	0.740
LR (1:6)	0.785	0.591	0.973	0.934	0.675	0.637	0.918	0.931	0.749
SVM (1:6)	0.841	0.550	0.983	0.929	0.665	0.641	0.921	0.915	0.762
RF (1:6)	0.799	0.615	0.974	0.938	0.695	0.659	0.923	0.940	0.776
ET (1:6)	0.816	0.600	0.977	0.936	0.692	0.659	0.925	0.943	0.779
DNN (1 layer, 1:1)	0.652	0.504	0.568	0.591	0.607	0.157	0.578	0.603	0.640
DNN (3 layers, 1:1)	0.737	0.497	0.823	0.620	0.593	0.338	0.659	0.637	0.692
DT (1:1)	0.802	0.791	0.805	0.794	0.797	0.596	0.798	0.798	0.849
NB (1:1)	0.836	0.708	0.862	0.747	0.767	0.576	0.785	0.874	0.872
KNN (1:1)	0.853	0.843	0.854	0.845	0.848	0.697	0.849	0.904	0.906
LR (1:1)	0.865	0.834	0.870	0.840	0.849	0.704	0.852	0.925	0.920
SVM (1:1)	0.855	0.858	0.854	0.858	0.857	0.713	0.856	0.928	0.921
RF (1:1)	0.844	0.886	0.836	0.880	0.864	0.723	0.861	0.932	0.920
ET (1:1)	0.853	0.879	0.849	0.876	0.866	0.729	0.864	0.934	0.928