Pandas 数据帧到Json的转换Python 方式

发布于12月29日

我对pandas到json的转换有点生疏了. 我有一个这样的pandas数据框，用于一个虚构的库:

userId	visitId	readingTime	Books	BookType
u1	1	300	book1,book2,book3	Fiction
u2	1	400	book4,book5	Horror
u2	2	250	book6	Romance

需要创建一个json，如下所示:

{
  "visitSummary": {
    "u1": [
      {
        "readingTime": 300,
        "Books": [
          "book1",
          "book2",
          "book3"
        ],
        "BookType": "Fiction"
      }
    ],
    "u2": [
      {
        "readingTime": 400,
        "Books": [
          "book4",
          "book5"
        ],
        "BookType": "Horror"
      },
      {
        "readingTime": 250,
        "Books": [
          "book6"
        ],
        "BookType": "Romance"
      }
    ]
  }
}

我正在考虑使用嵌套循环并处理每一行.我希望，有一种更简单的毒蛇疗法来做到这一点.

使用Python3.10和Pandas 2.1.4

推荐答案

split、groupby和to_dict:

out = {
    "visitSummary":
    (df.assign(Books=df["Books"].str.split(","))
       .groupby("userId").apply(lambda g: g.drop(
         columns=["userId", "visitId"]).to_dict("records")).to_dict())
}

发帖主题:Re:Kolibrios

import json; print(json.dumps(out, indent=4))
{
    "visitSummary": {
        "u1": [
            {
                "readingTime": 300,
                "Books": [
                    "book1",
                    "book2",
                    "book3"
                ],
                "BookType": "Fiction"
            }
        ],
        "u2": [
            {
                "readingTime": 400,
                "Books": [
                    "book4",
                    "book5"
                ],
                "BookType": "Horror"
            },
            {
                "readingTime": 250,
                "Books": [
                    "book6"
                ],
                "BookType": "Romance"
            }
        ]
    }
}