First 10 unique models | D1(n) | D1(%) | D2(n) | D2(%) | D3(n) | D3(%) | MPR(n) | MPR(%) | Comp (n) |
---|
M1a |
M1a | 425 | 88.0 | 402 | 83.2 | 410 | 84.9 | 441 | 91.3# | 483 |
M1b | 329 | 73.4 | 276 | 61.6 | 310 | 69.2 | 359 | 80.1# | 448 |
M1c | 178 | 49.7 | 146 | 40.8 | 194 | 54.2 | 270 | 75.4# | 358 |
M1d | 110 | 38.1 | 107 | 37.0 | 111 | 38.4 | 205 | 70.9# | 289 |
M2b |
M2a | 361 | 80.8 | 336 | 75.2 | 358 | 80.1 | 391 | 87.5# | 447 |
M2b | 253 | 67.3 | 231 | 61.4 | 250 | 66.5 | 280 | 74.5# | 376 |
M2c | 105 | 43.9 | 107 | 44.8 | 108 | 45.2 | 169 | 70.7# | 239 |
M2d | 93 | 52.8 | 94 | 53.4 | 93 | 52.8 | 109 | 61.9# | 176 |
M3c |
M3a | 491 | 98.2 | 489 | 97.8 | 491 | 98.2 | 492 | 98.4# | 500 |
M3b | 472 | 94.4 | 472 | 94.4 | 474 | 94.8 | 475 | 95.0# | 500 |
M3c | 452 | 90.4 | 445 | 89.0 | 455 | 91.0 | 456 | 91.2# | 500 |
M3d | 434 | 86.8 | 412 | 83.4 | 432 | 87.4 | 441 | 89.3# | 494 |
M4d |
M4a | 481 | 96.2 | 481 | 96.2 | 481 | 96.2 | 483 | 96.6# | 500 |
M4b | 439 | 88.5 | 465 | 93.8 | 469 | 94.6 | 470 | 94.8# | 496 |
M4c | 401 | 83.5 | 380 | 79.2 | 401 | 83.5 | 416 | 86.7# | 480 |
M4d | 93 | 52.8 | 94 | 53.4 | 77 | 43.8 | 112 | 63.6# | 176 |
- aM1 = model with n = 200. correlation degree 0.2; a = p-out ≤ 0.5; b = p-out ≤ 0.3; c = p-out ≤ 0.1; d = p-out t ≤ h0.05
- bM2 = model with n = 200. correlation degree 0.6; a = p-out ≤ 0.5; b = p-out ≤ 0.3; c = p-out ≤ 0.1; d = p-out ≤ 0.05
- cM3 = model with n = 500. correlation degree 0.2; a = p-out ≤ 0.5; b = p-out ≤ 0.3; c = p-out ≤ 0.1; d = p-out ≤ 0.05
- dM4 = model with n = 500. correlation degree 0.6; a = p-out ≤ 0.5; b = p-out ≤ 0.3; c = p-out ≤ 0.1; d = p-out ≤ 0.05
- n Number of observations, P-out P-value for excluding variable out of the model, D1 (n) Number of developed similar prognostic models as in the complete dataset with the D1-method, D1(%) Percentage of similar models as in the complete dataset with the D1-method, D2 (n) Number of developed similar prognostic models as in the complete dataset with the D2-method, D2(%) Percentage of similar prognostic models as in the complete dataset with the D2-method, D3 (n) Number of developed similar prognostic models as in the complete dataset with the D3-method, D3(%) Percentage of similar prognostic models as in the complete dataset with the D3-method, MPR (n) Number of developed similar prognostic models as in the complete dataset with the MPR-method, MPR (%) Percentage of similar prognostic models as in the complete dataset with the MPR-method, comp (n) Number of the first ten unique models selected in the BWS-procedure; # = highest amount of similar unique prognostic models compared to the models from the complete dataset