格式化熊猫数据框

Question

So I'm having an issue formatting in pandas. Any ideas? If this has been asked before please let me know and please point me to where! The return type doesn't matter to me as long as its formatted along the lines of what I have down there. I have the following where the numbers are the index in this case, New is my column.

   New
3  A
2  A
4  B
5  C
8  C

I want

A
3 2 
B
4 
C
5 8

Answer 1

It's a little bit different but you can nest a list as a value in a column inside a DataFrame in Python.

df = pd.DataFrame( {'new':['A','A','B','C','C'], 'val':[3,2,4,5,8]})
print(df)
     // apply a function 'list' on values and after
     // recreate a valid DataFrame where the field with lists
     // will be called 'up'

dfup = df.groupby('new')['val'].apply(list).reset_index(name='up')  
print("==============")
print(dfup)

The output is

  new  val
0   A    3
1   A    2
2   B    4
3   C    5
4   C    8
=============
  new      up
0   A  [3, 2]
1   B     [4]
2   C  [5, 8]

Normally, groupby is used to sum or some other aggregate function.

dfup = df.groupby('new').sum()

And the output is:

new
A      5
B      4
C     13

Answer 2

This is a one-liner with pivot_table() . You can use list as aggfunc :

> import pandas as pd
> df = pd.DataFrame({'New': ['A','A','B','C','C']}, index=[3,2,4,5,8])

> df.reset_index().pivot_table('index', aggfunc=list, columns='New')

New         A    B       C
index  [3, 2]  [4]  [5, 8]

# You probably want the transpose...

> df.reset_index().pivot_table('index', aggfunc=list, columns='New').T

      index
New        
A    [3, 2]
B       [4]
C    [5, 8]

This seems close enough to what you want, but you can add code to remove the brackets around the list, when formatted for print. (eg convert index to string, and use string concatenate or ' '.join() on it)