How to separate the groupedby data in diferent columns of a dataframe?

Question

How do i turn this

object	Name	Color
Fruit	Banana	Yellow
Fruit	Apple	Red
Fruit	Melon	Green
Car	Fiat	White
Car	BMW	Black
Car	NaN	NaN

In to this?

object	Name1	Name2	Name3	Color1	Color2	Color3
Fruit	Banana	Apple	Melon	Yellow	Red	Green
Car	Fiat	BMW	NaN	White	Black	NaN

I’ve searched the pandas documentation, but couldn’t find a solution to this

Read the pandas documentation, tried some diferent methods of groupby

Asked By: Gabriel Makhoul

||

Source

Answer 1

Feels inefficient, but you can first create a new column to keep track of the number of times each item is listed before melting, creating the new column names, then pivoting back.

import pandas as pd
import numpy as np

#original df
df = pd.DataFrame({
    'object': ['Fruit', 'Fruit', 'Fruit', 'Car', 'Car', 'Car'],
    'Name': ['Banana', 'Apple', 'Melon', 'Fiat', 'BMW', np.nan],
    'Color': ['Yellow', 'Red', 'Green', 'White', 'Black', np.nan],
})

#add an 'object_count' column to df
df['object_count'] = df.groupby('object').cumcount().add(1)

#melt df to long form
long_df = df.melt(id_vars=['object','object_count'])

#append 'object_count' to the variable column
long_df['variable'] += long_df['object_count'].astype(str)

#pivot the table back to wide form
final_df = long_df.pivot(
    index='object',
    columns='variable',
    values='value',
).reset_index()

final_df.columns.name = None #get rid of the 'variable' text at the top right of the table

#note, the output table isn't sorted by row or col the same as your expected output
#(it's sorted alphabetically for both)
#but you can do this or find help if it's important

print(final_df)

Output

  object  Color1 Color2 Color3   Name1  Name2  Name3
0    Car   White  Black    NaN    Fiat    BMW    NaN
1  Fruit  Yellow    Red  Green  Banana  Apple  Melon

Answered By: mitoRibo

Answer 2

With inspiration from comment by @mitoRibo, here is an answer:

df["N"] = df.assign(N=1).groupby("object")["N"].cumsum().map("Name{}".format)
df["C"] = df.assign(C=1).groupby("object")["C"].cumsum().map("Color{}".format)
out = df.pivot(index=["object"], columns=["N", "C"], values=["Name", "Color"])
out.columns = [t[1] if t[0] == "Name" else t[2] for t in out.columns]
print(out)

         Name1  Name2  Name3  Color1 Color2 Color3
object                                            
Car       Fiat    BMW    NaN   White  Black    NaN
Fruit   Banana  Apple  Melon  Yellow    Red  Green

Answered By: SomeDude

How to separate the groupedby data in diferent columns of a dataframe?

Question:

Answers: