1

I installed pandas-profiling with pip for jupyter notebook (not using conda!), and everything was working fine, until I installed plotly

Now, when I try to use the df.profile_report() method, I get the following error:

DispatchError: Function <code object pandas_missing_bar at 0x12c62e500, file "/Users/myuser/myvenv/lib/python3.9/site-packages/pandas_profiling/model/pandas/missing_pandas.py", line 18>

Please advise,

hortajg
  • 50
  • 6

1 Answers1

1

got the same error here.

This is what I did to solve this issue:

Uninstall all following dependencies and then install them again with these specific versions:

pandas-profiling==2.7.1
Jinja2==3.0.3
itsdangerous==2.0.1
Flask==1.1.1

Reference: https://github.com/pallets/flask/issues/4494

After doing that you may get this error while trying to generate a ProfileReport.

TypeError: Cannot interpret '<attribute 'dtype' of 'numpy.generic' objects>' as a data type

You can also solve this one by installing this specific version of numpy:

pip install numpy==1.16.5 --force-reinstall

Reference: https://stackoverflow.com/questions/66563463/how-to-treat-attribute-dtype-of-numpy-generic-objects-error

Worked for me, hope it helps you!

vitorrr
  • 46
  • 2
  • 1
    Cool! I ended up creating different venvs for profiling and for publication-grade plots, but this works! – hortajg Aug 01 '22 at 08:39