Table 2. Performance of classification model with variations in the r-value.

No	Resampling method	r	classifier	TPR (recall)	TNR	Precision (PPV)	FPR	AUC	Gmean	Adjusted Gmean	Fmeasure
1	Cluster-based only	1	DT	0.855a	0.769	0.454	0.231a	0.812	0.806	0.791	0.581
2	CluSMOTE	2	DT	0.797	0.834	0.526	0.163	0.815a	0.811a	0.823	0.622
3	CluSMOTE	3	DT	0.764	0.862	0.558	0.138	0.813	0.807	0.833	0.634
4	CluSMOTE	4	DT	0.730	0.881	0.575	0.119	0.806	0.796	0.835	0.631
5	CluSMOTE	5	DT	0.724	0.880	0.591	0.120	0.802	0.794	0.834	0.641
6	SMOTE only	–	DT	0.644	0.939a	0.732a	0.061	0.791	0.771	0.848a	0.675a
7	No Resampling	–	DT	0.637	0.939	0.730	0.061	0.788	0.767	0.846	0.669
8	Cluster-based only	1	SVM	0.591b	0.668	0.393	0.328b	0.629	0.579	0.620	0.388
9	CluSMOTE	2	SVM	0.577	0.746	0.441	0.254	0.661b	0.60b	0.666	0.400
10	CluSMOTE	3	SVM	0.498	0.790	0.486	0.210	0.644	0.580	0.675	0.396
11	CluSMOTE	4	SVM	0.475	0.801	0.508	0.199	0.638	0.566	0.672	0.387
12	CluSMOTE	5	SVM	0.468	0.819	0.529	0.178	0.643	0.572	0.683	0.401b
13	SMOTE only	–	SVM	0.384	0.881b	0.606b	0.119	0.632	0.532	0.688	0.368
14	No Resampling	–	SVM	0.409	0.874	0.569	0.126	0.641	0.557	0.699b	0.392

Notes.

TPR

True Positive Rate

TNR

True Negaitive Rate

AUC

Area Under ROC Curve

Gmean

Geometric mean

^aThe best parameter value in DT model.

^bThe best parameter vaue in SVM model.