FWHT-RF: A Novel Computational Approach to Predict Plant Protein-Protein Interactions via an Ensemble Learning Method

Table 4

Comparing results of RF with SVM and KNN model on three plants PPIs dataset.

Dataset	Classifier	Acc. (%)	Sen. (%)	Prec. (%)	MCC (%)	AUC (%)
Maize	RF	95.20 ± 0.38	92.99 ± 0.62	97.29 ± 0.26	90.85 ± 0.69	97.50 ± 0.33
	SVM	87.22 ± 0.41	86.26 ± 0.89	87.95 ± 0.71	77.70 ± 0.62	93.14 ± 0.44
	KNN	83.48 ± 0.38	91.29 ± 0.40	78.96 ± 0.87	72.08 ± 0.51	83.48 ± 0.30
Rice	RF	94.42 ± 0.56	94.17 ± 0.72	94.63 ± 0.84	89.46 ± 0.99	96.90 ± 0.37
	SVM	85.89 ± 0.91	86.65 ± 1.76	85.33 ± 0.55	75.76 ± 1.29	92.38 ± 0.49
	KNN	79.06 ± 0.65	88.86 ± 0.96	74.29 ± 0.91	66.25 ± 0.74	79.05 ± 0.55
Arabidopsis	RF	83.85 ± 0.35	76.95 ± 1.16	89.29 ± 0.62	72.66 ± 0.52	90.55 ± 0.41
	SVM	80.59 ± 0.37	77.22 ± 0.85	82.81 ± 0.41	68.65 ± 0.46	87.83 ± 0.45
	KNN	73.45 ± 0.41	78.53 ± 0.88	71.29 ± 0.72	60.79 ± 0.38	73.45 ± 0.40