6

I am trying to use numeric values as columns on a Pandas pivot_table. The problem is that since each number is mostly unique, the resulting pivot_table isn't very useful as a way to aggregate my data.

Here is what I have so far (fake data example):

import pandas as pd   

df = pd.DataFrame({'Country': ['US', 'Brazil', 'France', 'Germany'], 
                       'Continent': ['Americas', 'Americas', 'Europe', 'Europe'], 
                       'Population': [321, 207, 80, 66]})


pd.pivot_table(df, index='Continent', columns='Population', aggfunc='count')

Here is an image of the resulting pivot_table.

How could I group my values into ranges based on my columns?

In other words, how can I count all countries with Population... <100, 100-200, >300?

jlewkovich
  • 2,725
  • 2
  • 35
  • 49
Bruno Vieira
  • 151
  • 4
  • 9

1 Answers1

7

Use pd.cut:

df = df.assign(PopGroup=pd.cut(df.Population,bins=[0,100,200,300,np.inf],labels=['<100','100-200','200-300','>300']))

Output:

  Continent  Country  Population PopGroup
0  Americas       US         321     >300
1  Americas   Brazil         207  200-300
2    Europe   France          80     <100
3    Europe  Germany          66     <100

pd.pivot_table(df, index='Continent', columns='PopGroup',values=['Country'], aggfunc='count')

Output:

        Country          
PopGroup  200-300 <100 >300
Continent                  
Americas      1.0  NaN  1.0
Europe        NaN  2.0  NaN
Scott Boston
  • 147,308
  • 15
  • 139
  • 187