Table 5. One-dimensional analysis of parameters performed in DB10F.
| Classifier | Parameter |
 (%) |
 (%) |
 (%) |
| Bayes Net | -D | 0.00 | 0.00 | 0.00 |
| kNN | -K | 0.01 | 0.04 | 0.25 |
| kNN | -I | 0.00 | 0.00 | 0.00 |
| kNN | -F | 0.00 | 0.00 | 0.00 |
| kNN | -X | 0.00 | 0.00 | 0.00 |
| C4.5 | -U | −0.05 | 0.29 | 0.75 |
| C4.5 | -S | −0.01 | 0.13 | 0.25 |
| C4.5 | -A | 0.00 | 0.00 | 0.00 |
| C4.5 | -C | 0.27 | 0.30 | 1.25 |
| C4.5 | -M | 1.32 | 0.96 | 3.50 |
| C4.5 | -N | −7.44 | 1.75 | −2.75 |
| Logistic | -R | 0.58 | 0.71 | 4.25 |
| Logistic | -M | 0.81 | 0.73 | 4.25 |
| Naive Bayes | -K | −2.91 | 1.64 | 1.25 |
| Naive Bayes | -D | −19.20 | 3.10 | −11.75 |
| Perceptron | -D | −56.10 | 5.33 | −46.50 |
| Perceptron | -C | 0.00 | 0.00 | 0.00 |
| Perceptron | -H | 7.06 | 2.53 | 13.25 |
| Perceptron | -L | 2.27 | 1.11 | 5.50 |
| Perceptron | -M | 2.33 | 1.00 | 4.25 |
| Perceptron | -N | 1.01 | 0.77 | 4.00 |
| Perceptron | -V | 0.45 | 0.76 | 2.75 |
| Perceptron | -E | 0.00 | 0.00 | 0.00 |
| Random Forest | -I | 5.67 | 1.73 | 10.50 |
| Random Forest | -K | 0.54 | 0.98 | 3.75 |
| Random Forest | -depth | 1.11 | 1.03 | 3.75 |
| Random Forest | -S | 3.04 | 1.86 | 8.75 |
| Simple CART | -S | 1.09 | 0.81 | 2.75 |
| Simple CART | -C | 0.00 | 0.00 | 0.00 |
| Simple CART | -M | 1.41 | 1.16 | 4.25 |
| Simple CART | -N | 1.31 | 0.92 | 3.25 |
| Simple CART | -A | 3.89 | 1.70 | 9.00 |
| Simple CART | -H | 0.00 | 0.00 | 0.00 |
| Simple CART | -U | −1.21 | 1.18 | −4.00 |
| SVM | -C | 22.15 | 4.10 | 36.50 |
| SVM | -W | 0.38 | 0.33 | 1.50 |
| SVM | -P | 0.52 | 0.67 | 3.25 |
| SVM | -V | 0.00 | 0.00 | 0.00 |
| SVM | -N | 23.54 | 4.60 | 39.30 |
| SVM (poly kernel) | -E | 23.57 | 4.40 | 37.75 |
| SVM (NP kernel) | -E | 19.08 | 4.84 | 31.00 |
| SVM (RBF kernel) | -G | 21.55 | 3.91 | 34.75 |
| SVM (Puk kernel) | -S | 19.90 | 3.59 | 32.50 |
Comparison between the accuracy achieved with the default and the best parameter. The difference between the former and the latter in DB10F was summarized with the average, the standard deviation and the maximum difference. The appropriate tuning of C, N, E, G and S can improve significantly the performance of the default SVM.