Why I can't use groupby? Where is the problem?

Question

I have a large dataframe and I would like to use groupby function but I got the error:

TypeError: sequence item 0: expected str instance, list found

Data:

lokalnyid        object
powiat            int64
status_bdo       object
kategoria_       object
funkcja_og       object
funkcja_sz       object
zabytek           int64
geometry         object
liczba_kon      float64
1_5000_pi025     object
dtype: object

My code:

df2 = df.groupby(['lokalnyid', 'powiat', 'status_bdo', 'kategoria_', 'funkcja_og', 'funkcja_sz', 'zabytek', 'geometry', 'liczba_kon'])['1_5000_pi025'].apply(', '.join).reset_index()

I have read: Concatenate strings from several rows using Pandas groupby and I didn't find any solution.

Column joined_kafs_df_and_bud_layer['1_5000_pi025']:

0     [73634_913786_M-34-1-B-b-3-1.tif]
1    [73121_877737_M-34-41-D-b-3-1.tif]
2    [73710_920337_M-33-21-C-b-3-3.tif]
3     [72972_843871_M-34-2-B-d-2-4.tif]
4    [73864_954099_M-34-60-A-d-3-1.tif]
Name: 1_5000_pi025, dtype: object

What I want?

If some rows are the same according to 'lokalnyid', 'powiat', 'status_bdo', 'kategoria_', 'funkcja_og', 'funkcja_sz', 'zabytek', 'geometry', 'liczba_kon', group them and I would like to get lists in joined_kafs_df_and_bud_layer['1_5000_pi025'] like:

0     [73634_913786_M-34-1-B-b-3-1.tif, 73864_954099_M-34-60-A-d-3-1.tif]
 1    [73121_877737_M-34-41-D-b-3-1.tif, 73710_920337_M-33-21-C-b-3-3.tif]

After the `groupby` operation, why do you have `['1_5000_pi025']`? — DeGo, May 24 '21 at 11:24
Problem is you have `list`(s) in the `"1_5000_pi025"` column whereas they should individually be strings for `str.join` to work. — Mustafa Aydın, May 24 '21 at 11:28
Please share `1_5000_pi025` column's content and what exactly you wanted to see. You don't have to put the real data, an imitation is fine (better in fact). — Mustafa Aydın, May 24 '21 at 11:32
@datasciencebegginer So you are trying to concatanate the lists belonging the same group as I understand. Then, you can "sum" the lists: `df.groupby(...)["1_5000_pi025"].sum()`. Does that work? — Mustafa Aydın, May 24 '21 at 11:52

score 0 · Answer 1 · answered May 24 '21 at 12:57

0

You can .apply in groupby like this

df.groupby(['lokalnyid', 'powiat', 'status_bdo', 'kategoria_', 'funkcja_og', 'funkcja_sz', 'zabytek', 'geometry', 'liczba_kon'])['1_5000_pi025'].apply(list)

answered May 24 '21 at 12:57

Pawan Jain

815
3
15

this will result in lists like `[[3], [4], [10], [5]]` i.e. list-of-lists. – Mustafa Aydın May 24 '21 at 13:49

Why I can't use groupby? Where is the problem?

1 Answers1