python - 如何从 pandas 数据帧创建多级嵌套字典？

Question

我正在尝试从 pandas 数据帧创建一个多级嵌套字典 - 在下面的示例中，我想检索每个邮政编码、每个性别和年龄组合的工资总和。输出必须是预期输出注释中所示的字典。

from typing import NamedTuple, Sequence, Tuple

import pandas as pd

data = [
    ["tom", 22, "ab 11", "M", 5555],
    ["Rob", 22, "ab 11", "M", 9999],
    ["nick", 33, "ab 22", "M", 3333],
    ["juli", 18, "ab 11", "F", 2222],
]
people = pd.DataFrame(data, columns=["Name", "Age", "PostalCode", "Sex", "Salary"])

d = (
    people.groupby(["PostalCode", "Sex", "Age"])["Salary"]
    .apply(sum)
    .to_dict()
)

print(d)

# Expected output
print({"ab 11": {("M", 22): 15554, ("F", 18): 2222}, "ab 22": {("M", 33): 3333}})

Answer 1

只需稍微改变你的解决方案并使用额外的字典理解

df = (
    people.groupby(["PostalCode", "Sex", "Age"])["Salary"]
          .sum()
          .unstack(0)
    )

d =  {col: df[col].dropna().to_dict() for col in df}

print(d)

Out[40]:
{'ab 11': {('F', 18): 2222.0, ('M', 22): 15554.0},
 'ab 22': {('M', 33): 3333.0}}