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

Table 4

Performance of different multilabel classifiers based on features yielded by Node2vec under ten-fold cross-validation.

Scheme	Basic classification algorithm	Accuracy	Exact match	Hamming loss	Integrated score
RAKEL	Support vector machine (polynomial kernel)	0.774	0.698	0.050	0.513
	Support vector machine (RBF kernel)	0.738	0.668	0.059	0.464
	Random forest	0.762	0.695	0.054	0.501
Binary relevance	Support vector machine (polynomial kernel)	0.734	0.631	0.049	0.440
	Support vector machine (RBF kernel)	0.652	0.574	0.058	0.353
	Random forest	0.639	0.581	0.058	0.350