Table 7.
The overall accuracies (%) of 5-CV for three feature types and two hybrid feature types with SVM
| Dataset | Sen. | Spe. | MCC | PCPs | AAC | PSSM* | PCPs +AAC | PCPs +PSSM | |
|---|---|---|---|---|---|---|---|---|---|
| DNAset | Sa | 88.89 | 91.20 | 0.76 | 88.89 | 80.30 | 86.62 | 81.57 | 83.59 |
| Sr | 82.19 | 90.00 | 0.53 | 87.12 | 81.82 | 86.62 | |||
| DNAaest | Sa | 82.74 | 70.08 | 0.72 | 76.41 | 72.46 | 74.22 | 74.20 | 79.88 |
| Sr | 81.96 | 69.04 | 0.51 | 75.50 | 73.59 | 80.27 | |||
Sa: accurate solution, Sr: robust solution, Sen.: sensitivity, Spe.: specificity, MCC: Matthew’s correlation coefficient, PCPs: the m informative properties, PSSM*: obtained from [7].