Pandas groupby() merge different lists of strings

Question

I'm having the following dataframe.

Fruit	Description
Apple	["red", "big"]
Banana	["yellow", "long"]
Banana	["elongated, twisted"]
Peach	["round"]
Apple	["round", "greenish"]

And I'm trying to group by the descriptions according to the fruit, through a concatenation of the lists. I should obtain that:

Fruit	Description
Apple	["red", "big", "round", "greenish"]
Banana	["yellow", "long", "elongated, twisted"]
Peach	["round"]

I followed the solution provided here: pandas groupby and join lists:

df = df.groupby('Fruit', as_index=False).agg(Description =('Description', 'sum'))

but what I'm obtaining are lists attached to each other:

Fruit	Description
Apple	["red", "big"]["round", "greenish"]
Banana	["yellow", "long"]["elongated, twisted"]
Peach	["round"]

Does anyone have a solution? Thanks!

Answer 1

That's because your Description column is string. You can strip out the [] and sum:

 '[' + df['Description'].str[1:-1].groupby(df['Fruit']).agg(', '.join) + ']'

Answer 2

In order to keep your list format, I would suggest running a command before yours:

import json
df['Description'] = df['Description'].apply(json.loads)
df = df.groupby('Fruit', as_index=False).agg(Description =('Description', 'sum'))

That way, your values in the Description columns would be actual lists, and not strings.