熊猫分组汇总

Question

Let's say we have a pandas dataframe like the one below.

> category       level       score
>   Bus          travel      0.75
>   Bus          travel      0.60
>   Bus          vehicles    0.50

What I want is to group by the 'level' and calculate the 'count' and the maximum score for each 'level'. Also the 'hard' part is to create an output like this:

> category   travel  score    vehicles  score
>  Bus         2     0.75        1       0.5

I have been trying doing this:

>     grouped = df.groupby('level').agg(
    {
        'category': 'count',
        'score': 'max'
     })

Any ideas?

Answer 1

Setup

from StringIO import StringIO
import pandas as pd

text = """category       level       score
   Bus          travel      0.75
   Bus          travel      0.60
   Bus          vehicles    0.50"""

df = pd.read_csv(StringIO(text), delim_whitespace=1)

print df

  category     level  score
0      Bus    travel   0.75
1      Bus    travel   0.60
2      Bus  vehicles   0.50

Solution

gdf = df.groupby('category').apply(
    lambda df: df.groupby('level')['score'].agg({'count', 'max'})).unstack()

gdf.columns = gdf.columns.swaplevel(0, 1)
gdf = gdf.sort_index(axis=1)

print gdf

level    travel       vehicles     
          count   max    count  max
category                           
Bus           2  0.75        1  0.5