marah othman
08/14/2023, 4:30 PMCristian (Nixtla)
08/14/2023, 4:36 PMmarah othman
08/14/2023, 4:41 PMCristian (Nixtla)
08/14/2023, 4:42 PMmarah othman
08/14/2023, 4:42 PMCristian (Nixtla)
08/14/2023, 10:41 PMBradley de Leeuw
08/15/2023, 12:00 PMrefit_with_val=True
?Cristian (Nixtla)
08/15/2023, 3:15 PMrefit_with_val
argumentBradley de Leeuw
08/15/2023, 3:36 PMNeuralForecast.cross_validation
with the following settings:
h=15
step_size=15
n_windows=4
freq="W"
refit_with_val=True
So if I understand correctly, the last 60 weeks of data are never used to update the parameters of the model in this case? It simply trains on n - (4*15)
datapoints and only the input data changes over the 4 folds.Cristian (Nixtla)
08/15/2023, 4:43 PMrefit_with_val
is for training the best model afterwards. If true, it will include the validation set. But this is not part of the cross_validation
cross_validation
the model's parameter are not updatedval_size=0
if True. This is to fit the final model after selecting the best configuration.Bradley de Leeuw
08/16/2023, 8:40 AMMLForecast
has the refit=True
option in the cross_validation. Will this likely also be added for NeuralForecast
in the future?