Join same name dataframe column values into list

I create a dataframe with distinct column names and use rename to create columns with same name.

import pandas as pd

df = pd.DataFrame({
    "a_1": ["x", "x"],
    "a_2": ["", ""],
    "b_1": ["", ""],
    "a_3": ["", "y"],
    "c_1": ["z", "z"],
})

names = {
    "a_1": "a", 
    "a_2": "a", 
    "a_3": "a", 
    "b_1": "b", 
    "c_1": "c",
}

df2 = df.rename(columns=names)

This produces

   a a b  a  c
0  x         z
1  x      y  z

How to join values in the columns with same name into a list?

out = pd.DataFrame({
    "a": [["x"], ["x", "y"]],
    "b": [[""], [""]],
    "c": [["z"], ["z"]],
})

        a b    c
0     [x] [] [z]
1  [x, y] [] [z]

Bad attempt

I suspect this can be resolved with lambda

for c in ['a', 'b', 'c']:
    df2[c] = df2[c].apply(lambda x: x.tolist() if x.any() else [], axis=1)
    

This however produces

TypeError: <lambda>() got an unexpected keyword argument 'axis'

Any idea?

>Solution :

You could transpose and groupby.agg with a custom lambda:

df2.T.groupby(level=0).agg(lambda x: x[x!=''].tolist()).T

Output:

        a   b    c
0     [x]  []  [z]
1  [x, y]  []  [z]