Groupby.apply results are transposed when apply function returns Series

[gshimansky]

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04):

Ubuntu 20.04

• Modin version (modin.__version__):

0.7.3+167.gc685f41.dirty

• Python version:

Python 3.8.2

• Code we can use to reproduce:

import modin.pandas as pd
import numpy as np

df = pd.DataFrame(
    {
        "col1": ["a", "a", "a", "b", "b", "c", "c", "c"],
        "col2": [10, 20, 30, 40, 50, 60, 70, 80],
    })

def func(data):
    return pd.Series([111.11, 222.22])

print(df)
gb = df.groupby("col1")
print(gb.groups)
for col, frame in gb:
    print("Frame for col", col)
    print(frame)

df1 = gb.apply(func)
print(df1)
print(df1.shape)

Describe the problem

On Pandas there should be a DataFrame with contents like this

           0       1
col1
a     111.11  222.22
b     111.11  222.22
c     111.11  222.22

and shape (3, 2) but on Modin index goes into columns and Series contents appear in every cell:

col1                 a                 b                 c
col1
0     [111.11, 222.22]  [111.11, 222.22]  [111.11, 222.22]

This bug is related to #1587.

Source code / logs

[gshimansky]

Would it help to wrap user function in wrap_udf_function to cast Modin type to Pandas?

[mvashishtha]

@gshimansky On Modin version 0.11.1+9.g5e13fec, the code from your first comment produces a dataframe with shape (3, 2) as expected. I'm closing this issue.