Table 8. Average ranks of the seven classification algorithms.
| RBF | Lin | MLP | RF | KNN | DT | LVQ | |
|---|---|---|---|---|---|---|---|
| all | 2.38 | 2.83 | 2.93 | 3.39 | 4.07 | 5.94 | 6.45 |
| BD | 3.25 | 3.42 | 2.58 | 3.00 | 3.75 | 5.50 | 6.50 |
| CAD | 2.17 | 3.17 | 2.83 | 3.83 | 4.00 | 5.50 | 6.50 |
| CD | 2.83 | 3.17 | 2.83 | 3.17 | 3.00 | 6.50 | 6.50 |
| HT | 2.58 | 1.92 | 3.42 | 4.08 | 4.42 | 6.08 | 5.50 |
| RA | 2.50 | 2.83 | 2.50 | 3.50 | 3.67 | 6.50 | 6.50 |
| T1D | 1.50 | 3.17 | 3.00 | 3.00 | 5.00 | 5.33 | 7.00 |
| T2D | 1.83 | 2.17 | 3.33 | 3.17 | 4.67 | 6.17 | 6.67 |
The average ranks of the Friedman test for the seven different classifiers using the additive encoding. (Small values are better.) The result of the Friedman test over all data sets is significant (p < 10−15 for k = 7, n = 42). The table also shows the average ranks for each data set separately, but the Friedman test is not applicable here because the number of treatments is bigger than the number of problems (k = 7, n = 6).