CODEWITHSUNDEEP
pythonpython-2.7pandaspivot-table
owise
owise

Reputation: 1065

Converting long table to wide and creating columns according to the rows

I have a data frame that look like this:

Customer_ID     Category   Products 
  1               Veg         A
  2               Veg         B
  3              Fruit        A   
  3              Fruit        B
  3               Veg         B 
  1              Fruit        A
  3               Veg         C 
  1              Fruit        C

I want to find out the for each customer ID for each category which products were bought, and create a column for each product accordingly. The output would look like this:

Customer_ID     Category    Pro_1    Pro_2     Pro_3
  1               Veg        A        NA         NA
  1              Fruit       A        NA         C
  2               Veg        NA       B          NA
  3               Veg        NA       B          C
  3              Fruit       A        B          NA

Upvotes: 2

Views: 297

Answers (3)

jezrael
jezrael

Reputation: 862671

Use groupby with unstack, but if duplicates rows data are concanecate together:

df = df.groupby(['Customer_ID','Category','Products'])['Products'].sum().unstack()
df.columns = ['Pro_{}'.format(x) for x in range(1, len(df.columns)+1)]
df = df.reset_index()
print (df)
   Customer_ID Category Pro_1 Pro_2 Pro_3
0            1    Fruit     A  None     C
1            1      Veg     A  None  None
2            2      Veg  None     B  None
3            3    Fruit     A     B  None
4            3      Veg  None     B     C

Another solution with helper column, triples has to be unique:

#if not unique triples remove duplicates
df = df.drop_duplicates(['Customer_ID','Category','Products'])

df['a'] = df['Products']
df = df.set_index(['Customer_ID','Category','Products'])['a'].unstack()
df.columns = ['Pro_{}'.format(x) for x in range(1, len(df.columns)+1)]
df = df.reset_index()
print (df)
   Customer_ID Category Pro_1 Pro_2 Pro_3
0            1    Fruit     A  None     C
1            1      Veg     A  None  None
2            2      Veg  None     B  None
3            3    Fruit     A     B  None
4            3      Veg  None     B     C

Upvotes: 1

sheldonzy
sheldonzy

Reputation: 5961

Another option using crosstab :

pd.crosstab([df['Customer_ID'],df['Category']], df['Products'])

output:

Products              A  B  C
Customer_ID Category         
1           Fruit     1  0  1
            Veg       1  0  0
2           Veg       0  1  0
3           Fruit     1  1  0
            Veg       0  1  1

Afterwards you can reset the index for a similar solution to what you wanted.

df = df.reset_index()
Products  Customer_ID Category  A  B  C
0                   1    Fruit  1  0  1
1                   1      Veg  1  0  0
2                   2      Veg  0  1  0
3                   3    Fruit  1  1  0
4                   3      Veg  0  1  1

Upvotes: 1

Alex Ozerov
Alex Ozerov

Reputation: 1028

Try this: (don't mind the IO thing it is just for simple copy/paste)

import pandas as pd
from io import StringIO
df = pd.read_csv(StringIO("""
Customer_ID     Category   Products 
  1               Veg         A
  2               Veg         B
  3              Fruit        A   
  3              Fruit        B
  3               Veg         B 
  1              Fruit        A
  3               Veg         C 
  1              Fruit        C"""), sep='\s+')
df = df.join(pd.get_dummies(df['Products']))
g = df.groupby(['Customer_ID', 'Category']).sum()
print(g)

output:

                      A  B  C
Customer_ID Category         
1           Fruit     1  0  1
            Veg       1  0  0
2           Veg       0  1  0
3           Fruit     1  1  0
            Veg       0  1  1

Upvotes: 0

Related Questions