Fig. 8
From: Maximizing efficiency in sunflower breeding through historical data optimization
Summary of the methodology used in this work. From the seven years available, two have been selected as test sets (TS) and for each one the candidate sets (CS) can be selected manually of through optimization. For a given combination of TS-CS years, further optimization is possible to find the actual hybrids used in the training set (TRS), which is a subset of the CS. The TRS is subsequently used to train the gradient boosting machine (GBM) model employed to evaluate optimization performance. It is important to note that, when the TRS size is set to 100% of the CS, no optimization can take place, as the entire CS would be used as TRS