Table 3.

Model-2 selection results for severity categorization using traditional machine learning approach. Results reported are macro-averaged. Bold values indicated best results generated by feature weights and algorithms combination.

Dataset	Features Weights	Algorithm	Precison	Recall	F1-Score
		SVM_RBF	0.809	0.519	0.418
		SVM_Linear	0.792	0.698	0.705
	tp	Random_Forest	0.856	0.685	0.686
		Logistic_Regression	0.792	0.698	0.705
		KNeighbors	0.304	0.500	0.378
		SVM_RBF	0.797	0.655	0.649
		SVM_Linear	0.815	0.735	0.747
VHA	tf	Random_Forest	0.837	0.729	0.740
		Logistic_Regression	0.815	0.735	0.747
		KNeighbors	0.813	0.537	0.454
		SVM_RBF	0.720	0.562	0.512
		SVM_Linear	0.835	0.798	0.808
	tf-idf	Random_Forest	0.818	0.692	0.696
		Logistic_Regression	0.759	0.599	0.571
		KNeighbors	0.680	0.664	0.668
		SVM_RBF	0.458	0.500	0.478
		SVM_Linear	0.458	0.500	0.478
	tp	Random_Forest	0.458	0.500	0.478
		Logistic_Regression	0.458	0.500	0.478
		KNeighbors	0.458	0.500	0.478
		SVM_RBF	0.458	0.500	0.478
		SVM_Linear	0.460	0.500	0.475
VCU	tf	Random_Forest	0.458	0.478	0.478
		Logistic_Regression	0.460	0.490	0.473
		KNeighbors	0.458	0.500	0.478
		SVM_RBF	0.458	0.500	0.478
		SVM_Linear	0.460	0.495	0.475
	tf-idf	Random_Forest	0.458	0.500	0.478
		Logistic_Regression	0.458	0.500	0.478
		KNeighbors	0.457	0.490	0.473