I have tried this both on the hosted demo instance, as well as a locally deployed instance, but whenever I try the ‘gcp’ feature store example, I run into this error:
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-8-f062f5771128> in <module>
4 data = [['tom', 'johnson', 'UK'], ['penelope', 'charles', 'UK'], ['harry', 'windsor', "USA"]]
5 df = pd.DataFrame(data, columns=columns)
----> 6 name_country_fg.insert(df)
/srv/hops/anaconda/envs/theenv/lib/python3.7/site-packages/hsfs/feature_group.py in insert(self, features, overwrite, operation, storage, write_options)
706 operation,
707 storage.lower() if storage is not None else None,
--> 708 write_options,
709 )
710
/srv/hops/anaconda/envs/theenv/lib/python3.7/site-packages/hsfs/core/feature_group_engine.py in insert(self, feature_group, feature_dataframe, overwrite, operation, storage, write_options)
77 validation_id = None
78 if feature_group.validation_type != "NONE":
---> 79 validation = feature_group.validate(feature_dataframe)
80 validation_id = validation.validation_id
81
/srv/hops/anaconda/envs/theenv/lib/python3.7/site-packages/hsfs/feature_group.py in validate(self, dataframe)
943
944 """
--> 945 return self._data_validation_engine.validate(self, dataframe)
946
947 def get_validations(self, validation_time=None, commit_time=None):
/srv/hops/anaconda/envs/theenv/lib/python3.7/site-packages/hsfs/core/data_validation_engine.py in validate(self, feature_group, feature_dataframe)
50 expectations = self._expectations_api.get(feature_group=feature_group)
51
---> 52 expectation_results_java = engine.get_instance().validate(
53 feature_dataframe, expectations
54 )
AttributeError: 'Engine' object has no attribute 'validate'
To reproduce starting from the demo account, I created a new ‘feature store’ demo project (demo_fs_terence0) and started up a new JupyterLab instance. Then I selected ‘gcp’ folder, then ‘colab.ipnyb’. This works fine until about step 16 with the ‘name_country_fg.insert(df)’ command, which gives the above error.
I also started a new notebook on a new instance, and ran into a very similar issue attempting to create a feature store from scratch.