AttributeError: type object 'NDFrame' has no attribute 'groupby'

[gfranco008]

AttributeError: type object 'NDFrame' has no attribute 'groupby'

from pandas_ml import ConfusionMatrix
cm = ConfusionMatrix(actu, pred)
cm.print_stats()

---

AttributeError Traceback (most recent call last)
in
----> 1 from pandas_ml import confusion_matrix
2
3 cm = ConfusionMatrix(actu, pred)
4 cm.print_stats()

/usr/local/lib/python3.8/site-packages/pandas_ml/init.py in
1 #!/usr/bin/env python
2
----> 3 from pandas_ml.core import ModelFrame, ModelSeries # noqa
4 from pandas_ml.tools import info # noqa
5 from pandas_ml.version import version as version # noqa

/usr/local/lib/python3.8/site-packages/pandas_ml/core/init.py in
1 #!/usr/bin/env python
2
----> 3 from pandas_ml.core.frame import ModelFrame # noqa
4 from pandas_ml.core.series import ModelSeries # noqa

/usr/local/lib/python3.8/site-packages/pandas_ml/core/frame.py in
16 from pandas_ml.core.accessor import _AccessorMethods
17 from pandas_ml.core.generic import ModelPredictor, _shared_docs
---> 18 from pandas_ml.core.series import ModelSeries
19
20

/usr/local/lib/python3.8/site-packages/pandas_ml/core/series.py in
9
10
---> 11 class ModelSeries(ModelTransformer, pd.Series):
12 """
13 Wrapper for pandas.Series to support sklearn.preprocessing

/usr/local/lib/python3.8/site-packages/pandas_ml/core/series.py in ModelSeries()
78 return df
79
---> 80 @appender(pd.core.generic.NDFrame.groupby.doc)
81 def groupby(self, by=None, axis=0, level=None, as_index=True, sort=True,
82 group_keys=True, squeeze=False):

AttributeError: type object 'NDFrame' has no attribute 'groupby'

[joakimzhang]

any resolution？I also find this error

[Wenzkal]

I have the same problem as mentioned above

AttributeError: type object 'NDFrame' has no attribute 'groupby'.

Anyone facing this and have resolution...I am still trying

[alexjulien]

Same here, from the first
import pandas_ml as pdml
in a fresh pipenv with only pandas and pandas-ml

[elcolie]

Workaround. I put constraint like this and run with python3.7

pandas==0.24.2
pandas-ml
scikit-learn==0.20.0

[rtbs-dev]

This could be relevant: PR to remove (redundant?) NDFrameGroupBy. I don't thing the NDFrameGroupBy was intended as a generic interface that covered both the dataframe and series versions... the NDFrame definition does not implement a groupby method (though I weirdly see it being accessed here.

Seems like pandas-ml may want to use a different argument to the append decorator here. Based on digging through the style for DataFrame.groupby() and Series.groupby(), it seems the pandas devs already have a global _shared_docs object to use with their Appender decorator, so we should probably access this groupby docstring?

If that's an easy fix, whipping up a PR shouldn't be too bad. But I'd want to hear what @sinhrks has to say about it... any particular reason not to use the pandas.core._shared_docs variable in your Appender calls? Also might want to ask @WillAyd, since they seem to have deprecated this usage and I don't see it in the pandas v0.25 or v1.0.0 changelog? I'm just shooting in the dark here though. 🤷

[johnperry-math]

4 years later, this issue is still affecting at least some of us.

Fresh install of pandas_ml for Python 3.12.4.