KeyError: 'columns'

[leichtle]

Dear Eshin Jolly,
Running pymer4 on OS X with python 3.8.2, pandas 1.0.3, and rpy2 3.3.2 as well as R 4.0.0, lme4_1.1-23, and lmerTest_3.1-2, I can execute the linear modeling example without error:

# Import the linear regression model class
from pymer4.models import Lm

# Initialize model using 2 predictors and sample data
model = Lm('DV ~ IV1 + IV2', data=df)

# Fit it
print(model.fit())

However, the multilevel model throws an error:

# Import the lmm model class
from pymer4.models import Lmer

# Initialize model instance using 1 predictor with random intercepts and slopes
model = Lmer('DV ~ IV2 + (IV2|Group)', data=df)

# Fit it
print(model.fit())

This is the error message:

KeyError                                  Traceback (most recent call last)
~/python/lib/python3.8/site-packages/numpy/core/records.py in __getattribute__(self, attr)
    463         try:
--> 464             res = fielddict[attr][:2]
    465         except (TypeError, KeyError):

KeyError: 'columns'

During handling of the above exception, another exception occurred:

AttributeError                            Traceback (most recent call last)
~/python/lib/python3.8/site-packages/pymer4/models/Lmer.py in fit(self, conf_int, n_boot, factors, permute, ordered, summarize, verbose, REML, rank, rank_group, rank_exclude_cols, no_warnings, control, old_optimizer)
    403         try:
--> 404             self._get_ngrps()
    405         except AttributeError:

~/python/lib/python3.8/site-packages/pymer4/models/Lmer.py in _get_ngrps(self)
    228         res = get_ngrps(self.model_obj)
--> 229         if len(res.columns) != 1:
    230             raise Exception("Appears there's been another rpy2 or lme4 api change.")

~/python/lib/python3.8/site-packages/numpy/core/records.py in __getattribute__(self, attr)
    465         except (TypeError, KeyError):
--> 466             raise AttributeError("recarray has no attribute %s" % attr)
    467         obj = self.getfield(*res)

AttributeError: recarray has no attribute columns

During handling of the above exception, another exception occurred:

Exception                                 Traceback (most recent call last)
<ipython-input-30-59a6b50189ce> in <module>
      6 
      7 # Fit it
----> 8 print(model.fit())

~/python/lib/python3.8/site-packages/pymer4/models/Lmer.py in fit(self, conf_int, n_boot, factors, permute, ordered, summarize, verbose, REML, rank, rank_group, rank_exclude_cols, no_warnings, control, old_optimizer)
    404             self._get_ngrps()
    405         except AttributeError:
--> 406             raise Exception(
    407                 "You appear to have an old version of rpy2, upgrade to >3.0"
    408             )

Exception: You appear to have an old version of rpy2, upgrade to >3.0

Seems to me like a rpy2 version issue, maybe something for an updated pymer4 version.
Kind regards, Alex

[ejolly]

Yea sorry there are some compatibility issues with recent versions of rpy2 and pandas noted in #58 and #57. I haven't gotten a chance to work through them yet, but in the mean time you can trying using rpy2==3.2.4 and see if that works for you.

I also haven't tested this on the latest version of R so I can't guarantee it'll work on R >= 4.0

[ejolly]

@leichtle You can also try installing from the compatibility_fixes branch to see if that solves this particular issue. I think it should, but it may generate other ones since it's a WIP.

pip install git+https://github.com/ejolly/pymer4.git@compatibility_fixes

[ejolly]

Bug fixes

[ejolly]

Should not be an issue on upcoming releases, but add a slightly more informative error message