TABLE 2.
Classifier accuracy versus bootstrap confidence for the Bergey corpus
| Length of segment (bases) | % of correct classifier assignments within a bootstrap confidence range ofa:
|
|||||
|---|---|---|---|---|---|---|
| 100-95% | 94-90% | 89-80% | 79-70% | 69-60% | 59-50% | |
| Full | 98.0 | 66.4 | 69.2 | 41.8 | 46.2 | 34.7 |
| 400 | 98.3 | 86.1 | 75.9 | 65.4 | 61.1 | 49.2 |
| 200 | 98.2 | 90.1 | 83.0 | 75.6 | 64.6 | 55.7 |
| 100 | 97.4 | 89.8 | 82.5 | 75.6 | 64.7 | 55.6 |
| 50 | 94.9 | 83.9 | 76.8 | 67.9 | 59.5 | 49.7 |
a
Bootstrap confidence reflects the frequency of most common assignments out of 100 bootstrap samplings. Percentages of correct assignments at all ranks and within this bootstrap confidence range are shown.