Table 3 Performance of different classification models
input encoding | model | feature selection | average AUC (st. dev.) | average accuracy (st. dev.) | average TNR (st. dev.) | average TPR (st. dev.) |
|---|---|---|---|---|---|---|
bases | RF | none | 0.847 (0.100) | 83.948 (5.024) | 0.220 (0.230) | 0.992 (0.021) |
triplet | RF | none | 0.847 (0.097) | 83.952 (4.820) | 0.224 (0.222) | 0.990 (0.022) |
bases | RF | Fisher | 0.699 (0.160) * | 81.462 (5.812) | 0.215 (0.226) | 0.962 (0.055) |
triplet | RF | Fisher | 0.759 (0.127) | 81.781 (5.306) | 0.234 (0.219) | 0.961 (0.047) * |
bases | LR | Fisher/AIC | 0.670 (0.134) * | 81.310 (6.036) | 0.283 (0.220) | 0.943 (0.054) * |
triplet | LR | Fisher/AIC | 0.680 (0.147) * | 81.343 (6.878) | 0.324 (0.226) | 0.934 (0.058) * |
bases | DT | none | 0.570 (0.137) * | 79.862 (5.904) | 0.143 (0.217) | 0.960 (0.059) |
triplet | DT | none | 0.549 (0.106) * | 80.136 (4.850) * | 0.130 (0.203) | 0.967 (0.059) |
bases | RI | none | 0.574 (0.094) * | 80.662 (5.054) | 0.190 (0.219) | 0.958 (0.072) |
triplet | RI | none | 0.579 (0.110) * | 79.943 (5.442) | 0.215 (0.249) | 0.943 (0.074) |