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

Table 2

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

Scheme	Basic classification algorithm	Accuracy	Exact match	Hamming loss	Integrated score
RAKEL	Support vector machine (polynomial kernel)	0.818	0.754	0.042	0.591
	Support vector machine (RBF kernel)	0.757	0.670	0.055	0.479
	Random forest	0.803	0.743	0.045	0.570
Binary relevance	Support vector machine (polynomial kernel)	0.786	0.690	0.043	0.519
	Support vector machine (RBF kernel)	0.598	0.533	0.058	0.300
	Random forest	0.666	0.602	0.052	0.380