Table 2.
Preliminary results of different feature descriptors using different classifiers
| Features | Classifier (BACC%) | |||||
|---|---|---|---|---|---|---|
| LR | Bagging | RF | AB | GBM | SVM | |
| GPSD | 63.09 | 68.51 | 69.62 | 67.51 | 65.60 | 65.81 |
| ASDC | 58.53 | 63.13 | 66.88 | 62.96 | 64.03 | 64.51 |
| GAAPC | 58.91 | 66.75 | 67.71 | 64.44 | 67.56 | 61.75 |
| PAAC | 66.01 | 61.79 | 62.45 | 62.68 | 62.34 | 67.91 |
| GPSD_Ba | 67.12 | 75.23 | 79.62 | 73.54 | 76.91 | 76.13 |
| ASDC_Ba | 69.50 | 72.77 | 79.19 | 71.90 | 77.02 | 77.89 |
| GAAPC_Ba | 66.33 | 75.51 | 79.14 | 73.02 | 78.34 | 74.94 |
| PAAC_Ba | 66.63 | 71.77 | 77.46 | 69.91 | 74.29 | 73.09 |
The best performance value is highlighted in bold for clarification
aSMOTE-Tomek technique was applied to balance the data set