AttributeError: module 'numba' has no attribute 'generated_jit'

[melvinkokxw]

Current Behaviour

Trying to import ProfileReport (i.e. from ydata_profiling import ProfileReport) gives the error AttributeError: module 'numba' has no attribute 'generated_jit'.

Expected Behaviour

Import should happen without issues

Data Description

NA. Error is thrown before any dataset is used.

Code that reproduces the bug

from ydata_profiling import ProfileReport

pandas-profiling version

v4.7.0

Dependencies

ydata-profiling==4.7.0
numba==0.59.1

OS

macos

Checklist

• There is not yet another bug report for this issue in the issue tracker
• The problem is reproducible from this bug report. This guide can help to craft a minimal bug report.
• The issue has not been resolved by the entries listed under Common Issues.

[melvinkokxw]

generated_jit was removed in numba>=0.59.0. Temporary fix is to pin numba<=0.58.1.

[ebachura]

That works...unless you are on python 3.12, because you cannot install numba 0.58.1 on 3.12

[filipeo2-mck]

Setting minimal versions of numba and visions worked for me (for python 3.9):

# numba>=0.59.0 removed `numba.decorated_jit`
numba>=0.59.0
visions>=0.7.6

Edit: It looks like the above works if you are using pandas>=2.0. If pandas<2.0, these versions are working here:

numba<0.59.0
visions<0.7.6

[Sweetea99]

any luck on this one? I still have same issue

[muneer78]

Same here.

[jmakov]

And here. numba<=0.58.1 works as a tmp workaround for me.

[Keshu2311]

any permanent fix found? as none of the above fixes are working for me.

[jacquesfunk]

I had to create a venv using Python 3.11 to get it to work.

[adriens]

facing the same issue on Kaggle / Python 3.12, any hint how to patch the install section ?

# data analysis
!pip install --upgrade ydata-profiling
!pip install -U ydata-profiling[notebook,unicode]

[thierrymoudiki]

Anything new on this, please?

[thierrymoudiki]

And here. numba<=0.58.1 works as a tmp workaround for me.

This is a working temporary solution

[KansaiTraining]

Someone please post a complete workaround on this, since now the package is unusable