Fig 1.
Location of experimental sites [53].
Table 1.
Global structure of the machine learning modeling data sets.
Table 2.
Equations to compute climatic indices.
Table 3.
Variables used for modeling.
Table 4.
Description of trials used for model analysis.
Fig 2.
Predictive features importance for modeling.
Table 5.
Tuned model parameters.
Fig 3.
Comparison of models goodness of fit using R2, MAE and RMSE.
Fig 4.
Examples of potato yield response to N, P or K fertilization using different models.
Fig 5.
Examples of potato tuber size [M, S | L] balance response to N, P or K fertilization using different models.
Fig 6.
Examples of potato tuber size [S | M] balance response to N, P or K fertilization using different models.
Fig 7.
Examples of potato tuber SG response to N, P or K fertilization using different models.
Fig 8.
Economic or agronomic optimal doses and output predictions at optimal dosages for each model with a random selected test trial (N° 194).
Fig 9.
Examples of optimal economic N, P, K doses distribution with Gaussian processes using marketable yield for selected trials.
Fig 10.
Examples of agronomic optimal N, P, K doses distribution with Gaussian processes using tuber size [M, S | L] balance for selected trials.
Fig 11.
Examples of agronomic optimal N, P, K doses distribution with Gaussian processes using tuber size [S | M] balance for selected trials.
Fig 12.
Examples of agronomic optimal N, P, K doses distribution with Gaussian processes using tuber SG for selected trials.