Joint Nonnegative Matrix Factorization Based on Sparse and Graph Laplacian Regularization for Clustering and Co-Differential Expression Genes Analysis

Table 4

The contribution of graph regularization and sparse constraints to clustering performance.

	Datasets	Beta = 0 (%)	Lambda = 0, beta = 0 (%)
Accuracy	PAAD	49.70	78.87
	CHOL	1.82	5.68
	ESCA	22.52	23.93
	COAD	8.67	16.38
Recall	PAAD	15.90	26.22
	CHOL	56.77	58.98
	ESCA	−0.53	4.46
	COAD	14.30	13.15
Precision	PAAD	0.49	1.60
	CHOL	46.28	43.62
	ESCA	3.35	2.59
	COAD	6.18	4.92
F1-score	PAAD	15.67	13.72
	CHOL	53.26	53.68
	ESCA	0.70	3.78
	COAD	11.88	10.68