iMPTCE-Hnetwork: A Multilabel Classifier for Identifying Metabolic Pathway Types of Chemicals and Enzymes with a Heterogeneous Network

Table 3

Performance of different multilabel classifiers based on features yielded by DeepWalk under ten-fold cross-validation.

Scheme	Basic classification algorithm	Accuracy	Exact match	Hamming loss	Integrated score
RAKEL	Support vector machine (polynomial kernel)	0.333	0.290	0.145	0.083
	Support vector machine (RBF kernel)	0.342	0.304	0.140	0.089
	Random forest	0.334	0.297	0.142	0.085
Binary relevance	Support vector machine (polynomial kernel)	0.297	0.213	0.143	0.054
	Support vector machine (RBF kernel)	0.059	0.057	0.118	0.003
	Random forest	0.142	0.118	0.121	0.015