Fig. 1
From: LightGBM: accelerated genomically designed crop breeding through ensemble learning
Evaluation on the five basal machine learning (ML) models. a Data structure of the 42,840 F1s generated from the crossings of 1428 maternal lines and 30 paternal testers. The training samples include a total of 8652 F1s (207 maternal lines × 28 paternal testers, 1428 maternal lines × Zheng58, and 1428 maternal lines × Jing724). b Classification of the 6210 F1s by the OPTICS clustering algorithms. c Comparison of the five commonly used ML methods, namely support vector regression (SVR), random forests (RF), artificial neural network (ANN), k-nearest neighbor (KNN), and gradient boosting (GB) algorithms. GB exhibits precision and stability equivalent to that of rrBLUP