Can you describe the process of data processing in detail? about deepar-pytorch HOT 3 CLOSED

First you need to understand how a single time series works, here 370 time series together. Also if you could understand the following parameters:

window_size = 192
stride_size = 24
num_covariates = 4

pre-processing would be clear then.

from deepar-pytorch.

Thanks, I seem to understand, are these hyperparameters experimentally derived?
window_size means that each label consists of 192 consecutive data
stride_size means to skip 24 steps to take the next training data
num_covariates means that each data has four covariates to help predict

from deepar-pytorch.

Yes, You can perform you own tuning.

from deepar-pytorch.