1

I have the following dataset (non-unique id) :

id   data  country
1    8     B
2    15    A
3    14    D
3    19    D
3    8     C
3    20    A

For rows with country ANYTHING BUT "A", I want to add a "rank" column.

For rows with country "A", I want to leave "rank" value empty (or 0).

Expected output :

id   data  country rank
1    8     B       1
2    15    A       0
3    14    D       3 
3    19    D       4
3    8     C       2
3    20    A       0

This post Pandas rank by column value gives great insight.

I can try :

df['rank'] = df['data'].rank(ascending=True)

but I don't know how to take "country" into account ?

Community
  • 1
  • 1
Vincent
  • 1,534
  • 3
  • 20
  • 42

1 Answers1

5

EDIT: Written before an edit to the question so doesn't do exactly what the OP wants.

df['rank_A'] = df.data[df['country']=='A'].rank(ascending=True)

Tested on this 

 import pandas as pd
 from pandas import DataFrame
 import numpy as np
 df2 = DataFrame(np.random.randn(5, 2))
 df2.columns = ['A','B']
 df2['rank'] = df2.A[df2['B']>0].rank(ascending=True)
 df2

which gives the ranking according to A for rows in which B is greater than zero.

draco_alpine
  • 769
  • 11
  • 25