Error create a custom dataset

[StatMixedML]

Description

The following error is thrown (using the git install version) when I create a custom dataset following the tutorial. Seems to be a recent issue, as it as working Wednesday

To Reproduce

Using the extended forecasting tutorial see here

Error Message

---

GluonTSDataError Traceback (most recent call last)
in
58 feat_dynamic_real[:, :-custom_ds_metadata['prediction_length']],
59 feat_static_cat)],
---> 60 freq=custom_ds_metadata['freq'])

~/miniconda3/envs/gluonts/lib/python3.6/site-packages/gluonts/dataset/common.py in init(self, data_iter, freq, one_dim_target)
247 ) -> None:
248 process = ProcessDataEntry(freq, one_dim_target)
--> 249 self.list_data = [process(data) for data in data_iter]
250
251 def iter(self) -> Iterator[DataEntry]:

~/miniconda3/envs/gluonts/lib/python3.6/site-packages/gluonts/dataset/common.py in (.0)
247 ) -> None:
248 process = ProcessDataEntry(freq, one_dim_target)
--> 249 self.list_data = [process(data) for data in data_iter]
250
251 def iter(self) -> Iterator[DataEntry]:

~/miniconda3/envs/gluonts/lib/python3.6/site-packages/gluonts/dataset/common.py in call(self, data)
439 def call(self, data: DataEntry) -> DataEntry:
440 for t in self.trans:
--> 441 data = t(data)
442 return data
443

~/miniconda3/envs/gluonts/lib/python3.6/site-packages/gluonts/dataset/common.py in call(self, data)
378 if self.req_ndim != value.ndim:
379 raise GluonTSDataError(
--> 380 f"Array '{self.name}' has bad shape - expected "
381 f"{self.req_ndim} dimensions, got {value.ndim}."
382 )

GluonTSDataError: Array 'feat_dynamic_real' has bad shape - expected 2 dimensions, got 1.

Environment

• Operating system: Linux
• Python version: 3
• GluonTS version: gluonts-0.4.3.dev52+g6f0c16a

(Add as much information about your environment as possible, e.g. dependencies versions.)

[lostella]

Hi @StatMixedML the issue is probably due to the recently merged #574. The notebook you linked was later updated in #577 to reflect this change, see cells 24 and 25 in its latest version: can you confirm that this works?

Note that we added a link to the master ("latest") version of the documentation in the README, it's better to refer to that if you're working on the master branch, since the "stable" version doesn't get updated until a new release is done.

[StatMixedML]

Thanks @lostella, the example from the tutorial is working now. You can close the issue.