AWSMOTE: An SVM-Based Adaptive Weighted SMOTE for Class-Imbalance Learning

Table 5

Accuracy with different oversampling methods from real datasets.

Id	NO-RS	SMOTE	BLSMOTE	ADASYN	DBSMOTE	AWSMOTE
1	0.9691	0.9747	0.9626	0.9628	0.9668	0.9767
2	0.8962	0.9243	0.8911	0.9129	0.9135	0.9261
3	0.9162	0.9904	——	0.9953	0.9949	0.9973
4	0.9571	0.9648	——	0.9626	0.9709	0.9748
5	0.8489	0.9154	0.9371	0.9536	0.9164	0.9515
6	0.8985	0.9247	0.9215	0.9229	0.9016	0.9125
7	0.9133	0.9929	0.9855	0.9940	0.9877	0.9938
8	0.7212	0.8434	0.8814	0.9027	0.8494	0.9030
9	0.8942	0.9357	0.9385	0.9318	0.8900	0.9549
10	0.9640	0.9542	0.9731	0.9729	0.9768	0.9774
11	0.9220	0.9553	0.9543	0.9525	0.9490	0.9559
12	0.9602	0.9933	——	0.9837	0.9933	0.9943
13	0.9334	0.9862	——	0.9859	0.9867	0.9874
14	0.9534	0.9960	——	0.9952	0.9897	0.9978
15	0.9274	0.9765	——	0.9993	0.9988	1
16	0.9169	0.9939	——	0.9937	0.9939	0.9942
17	0.9118	0.9987	——	0.9985	0.9891	1
18	0.9808	0.9878	——	0.9926	0.9822	0.9935
19	0.9760	0.9798	——	0.9905	0.9878	0.9944
20	0.9523	——	——	——	0.9919	0.9960
21	0.9777	0.9795	——	0.9777	0.9905	0.9876
22	0.9428	0.9927	——	0.9943	0.9932	0.9946