Add multiple columns to dataframe: Object type error.

[iisri-vu]

System information

• **OS Platform and Distribution: Windows 10,
• Modin version (modin.__version__): 0.13.2
• Python version: Python 3.8
• Code we can use to reproduce:

cola = 1
colb = 2
colc = 5
columns= ['a', 'b', 'c']
df.loc[df['userid'].values == id, columns] = cola, colb, colc

Describe the problem

I get an error running the above code.
KeyError(array(['a'], dtype='<U6'))

Source code / logs

[mvashishtha]

@iisri-vu thank you for reporting the issue! I can't see how you initialize df in your example, but here's a small example that I think reproduces a similar error on the latest Modin version, 0.13.0+35.g47b8a1ce:

import modin.pandas as pd
df = pd.DataFrame({"a": [1]})
df.loc[:, "b"] = 2

I get a stack trace ending with KeyError: array(['b'], dtype='<U1'):

Show stack trace

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
Input In [58], in <module>
----> 1 df.loc[:, "b"] = 2

File ~/modin/modin/pandas/indexing.py:741, in _LocIndexer.__setitem__(self, key, item)
    739     self.qc = self.df._query_compiler
    740 else:
--> 741     row_lookup, col_lookup = self._compute_lookup(row_loc, col_loc)
    742     super(_LocIndexer, self).__setitem__(
    743         row_lookup,
    744         col_lookup,
   (...)
    748         ),
    749     )

File ~/modin/modin/pandas/indexing.py:868, in _LocIndexer._compute_lookup(self, row_loc, col_loc)
    858     if missing_mask.any():
    859         missing_labels = (
    860             # Converting `axis_loc` to maskable `np.array` to not fail
    861             # on masking non-maskable list-like
   (...)
    866             else axis_loc
    867         )
--> 868         raise KeyError(missing_labels)
    870 if isinstance(axis_lookup, pandas.Index) and not is_range_like(axis_lookup):
    871     axis_lookup = axis_lookup.values

KeyError: array(['b'], dtype='<U1')

whereas pandas will add the new column b so that the dataframe looks like:

   a  b
0  1  2

This is a Duplicate of #3764. Please follow that issue for the fix.

[mvashishtha]

Duplicate of #3764

[iisri-vu]

Thank you,
Do we have a nightly build version of Modin?

[mvashishtha]

No, there's no nightly build. You'll have to pull the latest source code from github after the fix.

[devin-petersohn]

@mvashishtha that's not completely accurate. @iisri-vu You can pip install directly from the latest github by following the instructions here: https://modin.readthedocs.io/en/stable/getting_started/installation.html#installing-from-the-github-master-branch

[iisri-vu]

There's still a bug in the newer version. We can't create multiple columns or add columns using this method.

[mvashishtha]

@iisri-vu yes, I'm sorry, but the bug #3764 is not fixed yet. I've only closed this issue so that we can track progress for both bugs in a single thread.

[iisri-vu]

Thanks for the fast response. This library will be a great one when we have these features in place.