Which Hyperparameters to tune? #527
Answered
by
ourownstory
ourownstory
asked this question in
Q&A - get help using NeuralProphet
-
|
Which hyperparameters should I leave on auto and which should I tune, if any? |
Beta Was this translation helpful? Give feedback.
Answered by
ourownstory
Jan 28, 2022
Replies: 1 comment
-
|
It depends. What are you forecasting and what are the data's characteristics? Without knowing about that, here are some quick pointers:
Hyperparameters should be tuned if hidden layers (NN used):
Further, if you have a panel dataset - you can train one global model with shared weights across all or train one model per time-series. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
ourownstory
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
It depends. What are you forecasting and what are the data's characteristics?
Without knowing about that, here are some quick pointers:
-- Y/N and N lags
-- Hidden layers: Y/N and N layers
Hyperparameters should be tuned if hidden layers (NN used):
Further, if you have a panel dataset - you can train one global model with shared weights across all or train one model per time-series.