Table 3.
for the five ML methods and for BLUP using a train-test split for the rice data. The best performance for each trait is in boldface. The average ranks for computation of the Friedman test are on the bottom line
| Trait/method | Lasso | Ridge | BLUP | GBM | RF | SVM |
|---|---|---|---|---|---|---|
| Culm diameter | 0.142 | 0.191 | 0.190 | 0.145 | 0.163 | 0.182 |
| Culm length | 0.529 | 0.567 | 0.568 | 0.559 | 0.539 | 0.516 |
| Culm number | 0.205 | 0.232 | 0.233 | 0.188 | 0.222 | 0.247 |
| Grain length | 0.387 | 0.375 | 0.381 | 0.371 | 0.361 | 0.383 |
| Grain width | 0.474 | 0.508 | 0.511 | 0.466 | 0.435 | 0.500 |
| Grain weight | 0.309 | 0.379 | 0.380 | 0.327 | 0.350 | 0.387 |
| Days to heading | 0.677 | 0.693 | 0.698 | 0.669 | 0.657 | 0.636 |
| Ligule length | 0.351 | 0.382 | 0.376 | 0.372 | 0.367 | 0.390 |
| Leaf length | 0.335 | 0.406 | 0.405 | 0.407 | 0.388 | 0.414 |
| Leaf width | 0.371 | 0.409 | 0.406 | 0.388 | 0.384 | 0.416 |
| Panicle length | 0.352 | 0.399 | 0.399 | 0.383 | 0.388 | 0.416 |
| Seedling height | 0.188 | 0.224 | 0.226 | 0.184 | 0.188 | 0.202 |
| Average rank | 4.88 | 2.38 | 2.12 | 4.50 | 4.71 | 2.42 |