Concat/Transpose/G​​roupby 熊猫专栏

Question

So I have the following DataFrame within pandas:

Column 1   | Column 2

Name       |  A
Number     |  B
Age        |  C
Name       |  D
Number     |  E
Age        |  F

Each Name, Number and Age grouped togther all relate to one feature and are repeated throughout the dateframe. I am and wondering what the best method would be to get it in the following format?:

           Name | Number | Age
Feature 1    A  |    B   |  C
Feature 2    D  |    E   |  F

Any help would be appreciated as I'm stumped as to what function or method I would use!

Answer 1

This is a pivot, but you first need to create a label to group the sets of 3 rows together. If the data are clean enough such that the DataFrame is always ordered Name, Number, Age, Name, Number, Age, ..., you can cumsum a Boolean Series checking which rows are 'Name' to group them together.

df['index'] = 'Feature ' + df['Column 1'].eq('Name').cumsum().astype(str)
#  Column 1 Column 2      index
#0     Name        A  Feature 1
#1   Number        B  Feature 1
#2      Age        C  Feature 1
#3     Name        D  Feature 2
#4   Number        E  Feature 2
#5      Age        F  Feature 2

df = (df.pivot(index='index', columns='Column 1', values='Column 2')
        .rename_axis(index=None, columns=None))
#          Age Name Number
#Feature 1   C    A      B
#Feature 2   F    D      E

---

Alternatively you could group every three rows together with integer division based on the length of the DataFrame.

import numpy as np

df['index'] = np.char.add(['Feature '], (np.arange(len(df))//3+1).astype(str))