error in function 'pd_time_unit' within 'transform_primitive.py'

[carter54]

should we delete the '.values' in 'getattr(pd_index, time_unit).values', otherwise I get the error of 'AttributeError: 'numpy.ndarray' object has no attribute 'values'' in this line.

def pd_time_unit(time_unit):
     def inner(pd_index):
           return getattr(pd_index, time_unit)
return inner  

[kmax12]

@carter54 can you past the full stack trace of the error? That will be helpful to see what primitive is causing this error, so we can try recreating it.

[carter54]

@kmax12 here are the full stack trace of the error ` File "", line 1, in runfile('D:/personal_code/Featuretools/simple_demo.py', wdir='D:/personal_code/Featuretools')

File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 866, in runfile execfile(filename, namespace)

File "C:\ProgramData\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 102, in execfile exec(compile(f.read(), filename, 'exec'), namespace)

File "D:/personal_code/Featuretools/simple_demo.py", line 11, in feature_matrix, features_defs = ft.dfs(entityset=es, target_entity="customers")

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\synthesis\dfs.py", line 186, in dfs verbose=verbose)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\calculate_feature_matrix.py", line 189, in calculate_feature_matrix target_time)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\calculate_feature_matrix.py", line 253, in calculate_batch _feature_matrix = calc_results(_time_last_to_calc, ids, precalculated_features=precalculated_features, training_window=window)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\calculate_feature_matrix.py", line 289, in wrapped r = method(*args, **kwargs)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\calculate_feature_matrix.py", line 231, in calc_results verbose=backend_verbose)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\pandas_backend.py", line 163, in calculate_all_features handler(group, entity_frames)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\computational_backends\pandas_backend.py", line 253, in _calculate_transform_features values = feature_func(*variable_data)

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\primitives\transform_primitive.py", line 216, in return lambda array: pd_time_unit(self.name)(pd.DatetimeIndex(array))

File "C:\ProgramData\Anaconda3\lib\site-packages\featuretools-0.1.16-py3.6.egg\featuretools\primitives\transform_primitive.py", line 534, in inner return getattr(pd_index, time_unit).values

AttributeError: 'numpy.ndarray' object has no attribute 'values'`

[kmax12]

@carter54 thanks. that is helpful.

We are trying to reproduce, but unable to.

Is there any chance you can share a piece of code that reproduces it? If you don't want to share publically, my gmail is in my profile.

[carter54]

@kmax12 I just followed the very simple example in the README.md

import featuretools as ft
es = ft.demo.load_mock_customer(return_entityset=True)
feature_matrix, features_defs = ft.dfs(entityset=es, target_entity="customers")
feature_matrix.head(5)

nothing else. When I delete the '.values' in 'getattr(pd_index, time_unit).values within 'transform_primitive.py'， everything works all right.

[kmax12]

I see. I am able to run that exact code using using python 3.6.4 and featuretools 0.1.16.

Perhaps you have a conflicting package installed. Can you run pip freeze and send what it outputs?

[kmax12]

Closing this for now. If you have additional information on how to reproduce, would be happy to help! Thanks

[SauceCat]

Hi @kmax12,

I encountered the same error when I tried to use training_window in dfs function and I fixed it by removing the '.values' in 'getattr(pd_index, time_unit).values'. :)

[Seth-Rothschild]

Thanks @SauceCat! I can now reproduce this with pandas version 0.19.2. We actually require pandas pandas>=0.20.3 in our requirements.txt. Is it possible something else that was installed after featuretools overwrote the pandas you were using?

Anyways, you should be able to fix by running pip install 'pandas>=0.20.3'.

[SauceCat]

@Seth-Rothschild yes, exactly. Thanks!