Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Model initialization fails with Python 3.8.0

See original GitHub issue

Hi, I’m not able to initialize an instance of Model() in the virtual environment I’m currently using:

Last updated: Tue Aug 03 2021

Python implementation: CPython
Python version       : 3.8.0
IPython version      : 7.26.0

matplotlib: 3.4.2
numpy     : 1.21.1
arviz     : 0.11.2
[GCC 8.4.0]
bambi     : 0.5.0
pandas    : 1.3.1

Watermark: 2.2.0

When initializing the instance, I get this error:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
/tmp/ipykernel_17456/2793899195.py in <module>
----> 1 model = Model(
      2     formula = 'measure_response_value ~ condition + (1 | wash_round)',
      3     priors = {'condition': 'wide'},
      4     data = data,
      5     dropna = True

~/mixed_models/bayesEnv/lib/python3.8/site-packages/bambi/models.py in __init__(self, formula, data, family, priors, link, categorical, dropna, auto_scale, default_priors, noncentered, taylor)
    162                         "Please rename the term(s) to prevent an unexpected behaviour."
    163                     )
--> 164             self._add_response(self._design.response, family, link, priors_)
    165         else:
    166             raise ValueError(

~/mixed_models/bayesEnv/lib/python3.8/site-packages/bambi/models.py in _add_response(self, response, family, link, priors)
    443                 )
    444 
--> 445         if response.refclass is not None and self.family.name != "bernoulli":
    446             raise ValueError("Index notation for response only available for 'bernoulli' family")
    447 

AttributeError: 'ResponseVector' object has no attribute 'refclass'

It looks weird since

I’m dealing with a continuous response variable
the error I’m getting back is not the same I see in the traceback

I tested the same analysis before in a local environment with these settings and it worked just fine:

Last updated: Tue Aug 03 2021

Python implementation: CPython
Python version       : 3.9.5
IPython version      : 7.22.0

matplotlib: 3.4.1
numpy     : 1.20.2
arviz     : 0.11.2
bambi     : 0.5.0
pandas    : 1.2.4

Watermark: 2.2.0

Is there an easy workaround for this? (Unfortunately I can’t change the Python version in the virtual environment I’m currently using).

Issue Analytics

State:
Created 2 years ago
Comments:8

Top GitHub Comments

1reaction

tomicaprettocommented, Aug 3, 2021

Great! Thanks for reporting the issue!

1reaction

arodavcommented, Aug 3, 2021

It works now with formulae==0.1.3, thanks!

Top Results From Across the Web

[tf2.0.0] Fails to build on Python 3.8 - suggested fix (change ...

ERROR : /home/daniel/tensorflow/tensorflow/python/BUILD:341:1: C++ ... 'std::nullptr_t' to 'Py_ssize_t {aka long int}' in initialization.

What's New In Python 3.8 — Python 3.11.1 documentation

To support both 3.8 and older, try python3-config --libs --embed first and fallback to python3-config --libs (without --embed ) if the previous command...

CPython DLL initialization routine failed from PYC cache file

While trying to use JPype on Windows Python 3.9.0, we are running into an bizarre issue with loading the internal module which is...

Python Initialization Configuration — Python 3.11.1 ...

New in version 3.8. Python can be initialized with Py_InitializeFromConfig() and the PyConfig structure.

Initialization, Finalization, and Threads — Python 3.11.1 ...

When the function returns, the current thread will hold the GIL and be able to call arbitrary Python code. Failure is a fatal...