pd.concat with Dictionary argument throws ValueError

[vidapura]

Hi,
Just installed modin for the first time.. and I'm failing at the first hurdle.
Could anyone please take a look and tell me if they can spot something obvious I'm doing wrong?
I'm using a macbook with 10.15.4 of the OS.
python 3

Installed modin using:

pip install modin[ray]

Now I'm trying to run code that worked with pandas:

import modin.pandas as pd
df_wave1 = pd.read_csv("Survey_1.csv")
df_wave2 = pd.read_csv("Survey_Cleaned_2.csv")
df_wave3 = pd.read_csv("Survey_Cleaned_3.csv")

surveys = {'1': df_wave1, '2': df_wave2, '3': df_wave3, }

df_all = pd.concat(surveys,  sort=False)

But I'm getting the error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-1-0ed75db51eaa> in <module>
      6 surveys = {'1': df_wave1, '2': df_wave2, '3': df_wave3, }
      7 
----> 8 df_all = pd.concat(surveys,  sort=False)

~/.virtualenvs/learningenv/lib/python3.7/site-packages/modin/pandas/concat.py in concat(objs, axis, join, ignore_index, keys, levels, names, verify_integrity, sort, copy)
     64             "and modin.pandas.DataFrame objs are "
     65             "valid",
---> 66             type(type_check),
     67         )
     68     all_series = all(isinstance(obj, Series) for obj in objs)

ValueError: ('cannot concatenate object of type "{0}"; only pandas.Series, pandas.DataFrame, and modin.pandas.DataFrame objs are valid', <class 'str'>)

I don't have pandas imported into the notebook.
I have restarted the kernel etc.

Any help would be greatly appreciated.
Thanks

[devin-petersohn]

Hi @vidapura, thanks for posting! This looks like a bug in Modin related to dictionary concatenation. We will get this fixed.

As a temporary workaround, you can slightly change your code to this:

import modin.pandas as pd
df_wave1 = pd.read_csv("Survey_1.csv")
df_wave2 = pd.read_csv("Survey_Cleaned_2.csv")
df_wave3 = pd.read_csv("Survey_Cleaned_3.csv")

# surveys = {'1': df_wave1, '2': df_wave2, '3': df_wave3, }
df_wave1.index = pd.MultiIndex.from_tuples([("1", i) for i in df_wave1.index])
df_wave2.index = pd.MultiIndex.from_tuples([("2", i) for i in df_wave2.index])
df_wave3.index = pd.MultiIndex.from_tuples([("3", i) for i in df_wave3.index])
surveys = [df_wave1, df_wave2, df_wave3]

df_all = pd.concat(surveys,  sort=False)

[vidapura]

Ahh brilliant!
Thanks @devin-petersohn !
I'll continue on with the experiment now and see if the rest of the code works with modin.
Thanks again.