Wrong numpy version?

[Vadim Peretokin]

• Cardea version: 0.1.2
• Python version: 3.8
• Operating System: Ubuntu 20.04 LTS

Description

Having just installed pip3 and cardea onto the system for the first time, the tutorial wasn't quite working for me:

python3 load.py 
Traceback (most recent call last):
  File "load.py", line 1, in <module>
    from cardea import Cardea
  File "/home/vadi/.local/lib/python3.8/site-packages/cardea/__init__.py", line 11, in <module>
    from cardea.core import Cardea
  File "/home/vadi/.local/lib/python3.8/site-packages/cardea/core.py", line 20, in <module>
    from cardea.modeling import Modeler
  File "/home/vadi/.local/lib/python3.8/site-packages/cardea/modeling/__init__.py", line 3, in <module>
    from cardea.modeling.modeler import Modeler
  File "/home/vadi/.local/lib/python3.8/site-packages/cardea/modeling/modeler.py", line 6, in <module>
    import sklearn
  File "/home/vadi/.local/lib/python3.8/site-packages/sklearn/__init__.py", line 76, in <module>
    from .base import clone
  File "/home/vadi/.local/lib/python3.8/site-packages/sklearn/base.py", line 16, in <module>
    from .utils import _IS_32BIT
  File "/home/vadi/.local/lib/python3.8/site-packages/sklearn/utils/__init__.py", line 15, in <module>
    from .murmurhash import murmurhash3_32
  File "__init__.pxd", line 242, in init sklearn.utils.murmurhash
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

Some googling around showed that installing latest numpy should do it - cardea installed numpy 1.18.5. I installed the latest one, 1.20.2, and that made it work.

[Vadim Peretokin]

I see that the next step unfortunately doesn't work:

Exception has occurred: ContextualVersionConflict
(numpy 1.20.2 (/home/vadi/.local/lib/python3.8/site-packages), Requirement.parse('numpy<1.19.0,>=1.16.0'), {'tensorflow'})
  File "/media/vadi/SSDer/Programs/cardea/load.py", line 6, in <module>
    label_times = cardea.select_problem('MissedAppointment')

So a version upgrade is not the right solution.

[Sarah Alnegheimish]

Hi @vadi2 thank you for opening this issue!

Cardea runs on python 3.6 and 3.7 (we are still working on supporting 3.8). Can you try creating a separate environment with python 3.6 for cardea? If you are using Anaconda you can run the following

conda create -n cardea python=3.6 # or 3.7

[Vadim Peretokin]

Thanks for the reply! I've created the Anaconda environment successfully (still new to this), but how to I install cardea? https://anaconda.org/search?q=cardea lists now results, and this doesn't work either:

conda search cardea --info
Loading channels: done
No match found for: cardea. Search: *cardea*

[Sarah Alnegheimish]

No problem!

you can directly use pip to download cardea, after you create the virtual environment, activate it using conda activate cardea from there use pip install cardea

[Vadim Peretokin]

Thanks! That got it going.