Table 2 Estimates of area under ROC curve (AUC) and 95% confidence intervals
From: Identifying genes that contribute most to good classification in microarrays
| Â | Â | Number of genes in classification rule | |||
|---|---|---|---|---|---|
Data set | Percent in training sample | 1 gene | 5 genes | 20 genes | 50 genes |
Colon cancer | 50% | .77 (.55 to .92) | .82 (.62 to .95) | .84 (.66 to .95) | .85 (.69 to .95) |
| Â | 80% | .86 (.59 to 1) | .90 (.69 to 1) | .89 (.69 to 1) | .90 (.70 to 1) |
Leukemia | 50% | .90 (.72 to 1) | .95 (.84 to .99) | .97 (.90 to .99) | .98 (.93 to 1) |
| Â | 80% | .95 (.76 to 1) | .97 (.84 to 1) | .99 (.91 to 1) | .99 (.93 to 1) |
Medulloblastoma | 50% | .60 (.50 to .77) | .62 (.50 to .78) | .65 (.50 to .79) | .67 (.51 to .82) |
| Â | 80% | .65 (.50 to .88) | .69 (.50 to .88) | .75 (.50 to .94) | .78 (.50 to .97) |
Breast Cancer | 50% | .81 (.58 to .94) | .88 (.71 to .97) | .91 (.75 to .99) | .91 (.76 to .99) |
| Â | 80% | .85 (.62 to 1) | .93 (.78 to 1) | .96 (.80 to 1) | .95 (.78 to 1) |