Research Article
The Construction of Support Vector Machine Classifier Using the Firefly Algorithm
Table 4
Classification results of the firefly-SVM and PSO-SVM with feature selection.
| Data sets | Number of original features | Firefly-SVM | PSO-SVM | CCR (mean ± SD) | Number of features | CCR (mean ± SD) | Average number of features |
| SPECTF heart | 44 | 85.89 ± 1.43 | 18 | 82.34 ± 2.174 | 22.6 | Breast Cancer Wisconsin (diagnosis) | 32 | 98.44 ± 0.39 | 14 | 98.44 ± 0.90 | 13.4 | Statlog (heart) | 13 | 86.75 ± 2.19 | 8 | 87.51 ± 2.21 | 8.6 | German credit data | 20 | 79.34 ± 1.14 | 12 | 75.33 ± 2.23 | 14.3 | Sonar | 60 | 95.23 ± 3.41 | 20 | 91.31 ± 2.42 | 32.7 | Pima-Indians-diabetes | 8 | 78.83 ± 1.62 | 5 | 77.87 ± 1.49 | 5.4 | Australian credit approval | 14 | 89.43 ± 1.65 | 8 | 84.34 ± 4.77 | 8.6 | Liver disorders | 7 | 75.36 ± 2.29 | 4 | 74.76 ± 2.12 | 5.2 | Ionosphere | 34 | 97.98 ± 3.15 | 12 | 95.21 ± 2.14 | 17.6 | Breast Cancer Wisconsin (original) | 10 | 98.67 ± 0.53 | 6 | 98.67 ± 1.39 | 6.7 |
|
|