Table 4.
ROC curves of simulation results conducted under different parameter settings where (EOO,woIF) setting is on are compared
| (EOO,woIF) | GAT(woGCM) | GLANET(woGCM) | GLANET(wGC) | GLANET(wM) | GLANET(wGCM) | Number of |
||
|---|---|---|---|---|---|---|---|---|
| Wins | Ties | Losses | ||||||
| GAT(woGCM) | 1/44/5 | 3/37/10 | 3/38/9 | 3/37/10 | 10 | 156 | 34 | |
| GLANET(woGCM) | 5/44/1 | 3/38/9 | 3/38/9 | 3/38/9 | 14 | 158 | 28 | |
| GLANET(wGC) | 10/37/3 | 9/38/3 | 5/41/4 | 3/43/4 | 27 | 159 | 14 | |
| GLANET(wM) | 9/38/3 | 9/38/3 | 4/41/5 | 3/42/5 | 25 | 159 | 16 | |
| GLANET(wGCM) | 10/37/3 | 9/38/3 | 4/43/3 | 5/42/3 | 28 | 160 | 12 | |
*
Note: A win indicates a case where the ROC curve obtained with settings specified in the row is statistically significantly above the ROC curve obtained with the settings specified in the column at significance level of 0.05. A loss indicates the opposite, while a tie indicates that there is no statistically significant difference between the two compared curves. The counts indicate the number of times win/tie/loss cases occur when the results for different elements, cell lines and other experimental conditions are compared. The best parameter setting is shown in bold.