Bivariate distribution dependency

[shakasom]

Plotting bivariate distributions with scatter plot combined with box/violin plots throws dependency error.

ModuleNotFoundError: No module named 'patsy'

Environment: Windows 10
Statsmodels: 0.9.0
Scipy: 1.1.0
pandas: 0.24.2
Plotly: 3.7.1

[empet]

@shakasom I had the same issue, but installing statsmodels for python 3.7.1 fixed it.

conda install -c conda-forge statsmodels

performed the following replacement:
0.9.0-py36hfa6e2cd_1000 --> 0.9.0-py37hfa6e2cd_1000

and upgraded patsy to patsy-0.5.1

[nicolaskruchten]

Ah right, I forget that installing statsmodels doesn't always install its dependencies. I'll add this stuff to setup.py tomorrow and push to PyPI again :)

Thanks for letting me know!

[shakasom]

Thanks @empet I installed patsy with pip and everything worked.