Question

我有以下pandas dataframe

+---------+-------+
| Country | value |
+---------+-------+
| UK      |    42 |
| US      |     9 |
| US      |    10 |
| France  |    15 |
| France  |    16 |
| Germany |    17 |
| Germany |    18 |
| Germany |    20 |
+---------+-------+

我想创建一个新列，根据从最大到最小的值的平均值对每个国家/地区进行排名

输出如下所示

+---------+-------+---------+------+
| Country | value | Average | Rank |
+---------+-------+---------+------+
| UK      |    42 |      42 |    1 |
| US      |     9 |     9.5 |    4 |
| US      |    10 |     9.5 |    4 |
| France  |    15 |    15.5 |    3 |
| France  |    16 |    15.5 |    3 |
| Germany |    17 |      18 |    2 |
| Germany |    18 |      18 |    2 |
| Germany |    20 |      18 |    2 |
+---------+-------+---------+------+

请注意，我不需要平均列，它只是帮助解释。

非常感谢

Answer 1

对mean使用groupby + transform，然后使用rank：

df['Average'] = df.groupby('Country')['value'].transform('mean')
df['Rank'] = df['Average'].rank(method='dense', ascending=False)
print (df)
   Country  value    Average  Rank
0       UK     42  42.000000   1.0
1       US      9   9.500000   4.0
2       US     10   9.500000   4.0
3   France     15  15.500000   3.0
4   France     16  15.500000   3.0
5  Germany     17  18.333333   2.0
6  Germany     18  18.333333   2.0
7  Germany     20  18.333333   2.0

类似的解决方案：

a = df.groupby('Country')['value'].transform('mean')
b = a.rank(method='dense', ascending=False)

df = df.assign(Average=a, Rank=b)
print (df)
   Country  value    Average  Rank
0       UK     42  42.000000   1.0
1       US      9   9.500000   4.0
2       US     10   9.500000   4.0
3   France     15  15.500000   3.0
4   France     16  15.500000   3.0
5  Germany     17  18.333333   2.0
6  Germany     18  18.333333   2.0
7  Germany     20  18.333333   2.0

Answer 2

<强>解决方案
在pd.DataFrame.join与pd.concat

之后我使用了groupby和mean的组合

m = df.groupby('Country').value.mean()
df.join(
    pd.concat([m, m.rank(ascending=False)], axis=1, keys=['Average', 'Rank']),
    on='Country')

   Country  value    Average  Rank
0       UK     42  42.000000   1.0
1       US      9   9.500000   4.0
2       US     10   9.500000   4.0
3   France     15  15.500000   3.0
4   France     16  15.500000   3.0
5  Germany     17  18.333333   2.0
6  Germany     18  18.333333   2.0
7  Germany     20  18.333333   2.0

同样，使用双join

m = df.groupby('Country').value.mean()
df.join(m.rename('Avergage'), on='Country') \
  .join(m.rank(ascending=False).rename('Rank'), on='Country')

   Country  value    Average  Rank
0       UK     42  42.000000   1.0
1       US      9   9.500000   4.0
2       US     10   9.500000   4.0
3   France     15  15.500000   3.0
4   France     16  15.500000   3.0
5  Germany     17  18.333333   2.0
6  Germany     18  18.333333   2.0
7  Germany     20  18.333333   2.0

或map和assign

m = df.groupby('Country').value.mean()
df.assign(
    Average=df.Country.map(m),
    Rank=df.Country.map(m.rank(ascending=False))
)

   Country  value    Average  Rank
0       UK     42  42.000000   1.0
1       US      9   9.500000   4.0
2       US     10   9.500000   4.0
3   France     15  15.500000   3.0
4   France     16  15.500000   3.0
5  Germany     17  18.333333   2.0
6  Germany     18  18.333333   2.0
7  Germany     20  18.333333   2.0

Answer 3

我使用现代方法链接方法来避免变异状态和创建新变量：

df = pd.DataFrame(
    {'Country': ['Russia', 'Russia', 'USA'], 'Value': [12, 15, 16]})

df.join(df.groupby('Country').
           mean().
           rank().
           rename(columns={'Value': 'Rank'}),
        on='Country')

在pandas中，如何根据另一列的平均值创建一个具有排名的新列

3 个答案: