Table 3.

The average RMSE of ten simulations obtained with the GRU model, for the two datasets and two sizes of the recurrent layer (4 and 8 hidden units).

	Dataset 1–4	Dataset 2–4	Dataset 1–8	Dataset 2–8
Training	1.05e−02	1.08e−02	7.79e−03	8.17e−03
Validation	3.48e−02	3.45e−02	3.47e−02	3.54e−02
Test	1.17e−02	1.21e−02	1.00e−02	1.04e−02