CODEWITHSUNDEEP
pythonpandas
snapcrack
snapcrack

Reputation: 1811

pandas groupby rank removes index, returns all 1s

My dataframe looks like this:

    name1  name2   value
 1  Jane   Foo     2
 2  Jane   Bar     4
 3  John   Foo     7
 4  John   Bar     1

If I do df.groupby(['name1', 'name2']).count() I get:

              value
name1  name2
Jane   Foo    1
Jane   Bar    1
John   Foo    1
John   Bar    1

But I'm trying to find the rank of each value within each multiindex group. Ideally, if I use df.groupby(['name1', 'name2']).rank() I should get:

             value
name1  name2
Jane   Foo    2
Jane   Bar    1
John   Foo    1
John   Bar    2

But instead I simply get:

    value
  1  1
  2  1
  3  1
  4  1

with the names of the grouped columns removed, only the index numbers as the index, and the rank value for all rows equaling 1. What am I doing wrong?

Upvotes: 1

Views: 482

Answers (1)

jezrael
jezrael

Reputation: 862511

I think you need working with numeric - so it seems need grouping be first column name1 and return rank for value:

df['rank'] = df.groupby('name1')['value'].rank(method='dense', ascending=False).astype(int)
print (df)
  name1 name2  value  rank
1  Jane   Foo      2     2
2  Jane   Bar      4     1
3  John   Foo      7     1
4  John   Bar      1     2

Upvotes: 4

Related Questions