business-science / modeltime.gluonts

GluonTS Deep Learning with Modeltime
https://business-science.github.io/modeltime.gluonts/
Other
39 stars 9 forks source link

Help - Error in py_call_impl -> Array 'target' has bad shape - deepar or nbeats #48

Open sextant96 opened 2 years ago

sextant96 commented 2 years ago

Hello Modeltime GluonTS Team:

I kindly ask your help or guidance to solve this error that isn't allowing me to build deepar or nbeats forecast models according to your doc https://business-science.github.io/modeltime.gluonts/articles/managing-envs.html and https://business-science.github.io/modeltime.gluonts/reference/deep_ar.html:

1) I'm getting this error trying to build deepar or nbeats models using your fit function on any of these fit modes, with gluonts_nbeats_ensemble or gluonts_deepar engine in each case:

fit(value ~ ., training())

fit(value ~ date + id, )

Error in py_call_impl(callable, dots$args, dots$keywords) : gluonts.core.exception.GluonTSDataError: Array 'target' has bad shape - expected 1 dimensions, got

2) I fixed a similar "regression" error, reinstalling all your modeltime gluonts enviroment, including conda and all necessary visual studio build tools required packages as instructed on this link: https://developercommunity.visualstudio.com/t/error-microsoft-visual-c-140-is-required/409173

3) I also added the required id column on and with rowid_to_column function.

4) As I saw the error had a py call, I tried to install and reinstall the install_gluonts - include_pytorch option on either true or false flag, but both ways return the same error.

Regards.

sextant96 commented 2 years ago

Hello Modeltime GluonTS Team:

1) Reviewing your youtube channel's videos, on this https://www.youtube.com/watch?v=QhuoCKhQ5fg min 34:13, code lines 111-112, I tried to replicate your code for fit command with "data", still without avail (appears that bad shape array error).

Regards.