Table 3 Evaluation of proposed method on Iris dataset in terms of accuracy (%) with tenfold cross validation (best results denote in italics typeface)
Classifier | #Prototype | ||||
|---|---|---|---|---|---|
1 | 20 | 75 | 100 | 149 | |
1-nn | 96.00(± 1.2) | 95.33(± 0.8) | 96.00(± 0.2) | 96.00(± 0.8) | 96.00(± 0.7) |
3-nn | 94.00(± 2.1) | 95.33(± 1.1) | 94.00(± 1.2) | 96.67(± 1.6) | 96.00(± 1.4) |
5-nn | 95.33(± 2.3) | 95.33(± 1.3) | 93.33(± 2.4) | 96.67(± 1.4) | 95.33(± 1.2) |
7-nn | 91.33(± 3.3) | 96.00(± 1.1) | 92.67(± 3.2) | 96.00(± 1.9) | 97.33(± 0.5) |
11-nn | 94.00(± 2.5) | 96.00(± 0.9) | 94.67(± 2.1) | 96.67(± 1.3) | 96.67(± 1.2) |
linSVM | 94.67(± 1.1) | 98.00(± 0.2) | 95.33(± 2.3) | 95.33(± 2.3) | 95.33(± 2.3) |
polySVM(2) | 94.67(± 1.3) | 97.33(± 0.8) | 95.33(± 1.2) | 94.67(± 3.1) | 95.33(± 1.2) |
polySVM(3) | 94.67(± 2.3) | 97.33(± 1.5) | 95.33(± 1.4) | 94.00(± 3.2) | 94.67(± 1.3) |
Average | 94.33(± 1.38) | 96.33(± 1.07) | 94.58(± 1.15) | 95.75(± 1.0) | 95.83(± 0.85) |