Table 3. Assessment of the predictive power of the seven biometrical models by fivefold cross-validation with 1000 runs.
| Model | A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|---|
| BY | |||||||
| QTL | 13.7 | 9.8 | 9.4 | 31.9 | 31.5 | 3.3 | 4.6 |
| pG−ES | 73.1 | 52.1 | 59.0 | 64.9 | 66.1 | 35.5 | 35.0 |
| pG−TS | 49.9 | 39.6 | 35.6 | 51.9 | 50.8 | 31.8 | 33.8 |
| Bias | 31.7 | 24.0 | 39.7 | 20.0 | 23.1 | 10.4 | 3.4 |
| K | |||||||
| QTL | 12.4 | 10.6 | 7.2 | 11.1 | 11.3 | 3.3 | 2.2 |
| pG−ES | 70.8 | 52.6 | 26.6 | 52.6 | 58.0 | 36.3 | 24.2 |
| pG−TS | 57.5 | 41.3 | 15.2 | 43.1 | 48.1 | 29.4 | 15.8 |
| Bias | 18.8 | 21.5 | 42.9 | 18.1 | 17.1 | 19.0 | 34.7 |
QTL, average number of main-effect QTL detected in the estimation set; pG−ES, the proportion of explained genotypic variance (%) of these QTL in the estimation set and in the test set (pG−TS), and the relative bias in the proportion of explained genotypic variance for beet yield (BY) and potassium content (K).