将具有多个索引的数据框转换为具有列值的键的键

Question

我的项目的用例是显示每个作者的提交以及每天每个提交的大小。这就是我需要表示我的数据的方式

为此，我所做的是

timed_commits = commit_data.set_index('Date')
grouped = timed_commits.groupby(by=["Author"])
resampled = grouped.resample("D").agg(
            {"SHA": "size", "Insertion": "sum", "Deletion": "sum"}
        ) # get the total count of commits in a day with total insertion and deletion

这给了我下面的输出

这里作者和日期是索引，而 SHA、插入和删除是列。 Author 和 Date 是索引的原因是我想知道每个作者每天的提交，同时我也想知道每个提交的大小（通过插入）

对于这样的对象，我无法以这种方式或任何其他方式格式化（最好有作者值和日期值的字段名称），这将支持在表格中显示，如我为它附上的图片

{
    'author1': {
        '2017-10-18': {'SHA': 1, 'Insertion': 1.0, 'Deletion': 3.0},
        '2017-10-19': {'SHA': 2, 'Insertion': 1.0, 'Deletion': 3.0},
        '2017-10-20': {'SHA': 6, 'Insertion': 1.0, 'Deletion': 3.0},
        '2017-10-21': {'SHA': 9, 'Insertion': 1.0, 'Deletion': 3.0},
    },
    'author2': {
        '2017-10-18': {'SHA': 3, 'Insertion': 8.0, 'Deletion': 3.0},
        '2017-10-19': {'SHA': 19, 'Insertion': 10.0, 'Deletion': 3.0},
        '2017-10-20': {'SHA': 23, 'Insertion': 1.0, 'Deletion': 3.0},
        '2017-10-21': {'SHA': 44, 'Insertion': 1.0, 'Deletion': 3.0},
    }
}

我玩过 to_dict，但没有让它可行。

这是数据帧（这里的提交哈希，即 sha 由于该特定提交中更改的文件数量而重复）。这是从 git 日志中获取的。

SHA Timestamp   Date    Author  Insertion   Deletion    Churn   File path
1   cae635054   Sat Jun 26 14:51:23 2021 -0400  2021-06-26 18:51:23+00:00   Andrew Clark    31.0    0.0 31.0    packages/react-reconciler/src/__tests__/ReactI...
2   cae635054   Sat Jun 26 14:51:23 2021 -0400  2021-06-26 18:51:23+00:00   Andrew Clark    1.0 1.0 0.0 packages/react-test-renderer/src/ReactTestRend...
3   cae635054   Sat Jun 26 14:51:23 2021 -0400  2021-06-26 18:51:23+00:00   Andrew Clark    24.0    14.0    10.0    packages/react/src/ReactAct.js
5   e2453e200   Fri Jun 25 15:39:46 2021 -0400  2021-06-25 19:39:46+00:00   Andrew Clark    50.0    0.0 50.0    packages/react-reconciler/src/__tests__/ReactI...
7   73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    4.0 5.0 -1.0    packages/react-devtools-shared/src/__tests__/F...
8   73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    4.0 4.0 0.0 packages/react-devtools-shared/src/__tests__/c...
9   73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    12.0    12.0    0.0 packages/react-devtools-shared/src/__tests__/c...
10  73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    7.0 6.0 1.0 packages/react-devtools-shared/src/__tests__/e...
11  73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    47.0    42.0    5.0 packages/react-devtools-shared/src/__tests__/i...
12  73ffce1b6   Thu Jun 24 22:42:44 2021 -0400  2021-06-25 02:42:44+00:00   Brian Vaughn    7.0 6.0 1.0 packages/react-devtools-shared/src/__tests__/o...

Answer 1

也许我完全误解了。

我假设您要转换 DataFrame df

                    SHA  Insertion  Deletion
Author  Date                                
author1 2017-10-18    1        1.0       3.0
        2017-10-19    2        1.0       3.0
        2017-10-20    6        1.0       3.0
        2017-10-21    9        1.0       3.0
author2 2017-10-18    3        8.0       3.0
        2017-10-19   19       10.0       3.0
        2017-10-20   23        1.0       3.0
        2017-10-21   44        1.0       3.0

进入您提供的dict-格式？

如果是这样，那么试试这个：

result = {
    key: group.reset_index(level=0, drop=True).to_dict(orient='index')
    for key, group in df.groupby('Author')
}

或者这个

result = (df.groupby('Author')
            .apply(lambda sdf: sdf.reset_index(level=0, drop=True).to_dict(orient='index'))
            .to_dict())

样本结果：

{'author1': {'2017-10-18': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 1},
             '2017-10-19': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 2},
             '2017-10-20': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 6},
             '2017-10-21': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 9}},
 'author2': {'2017-10-18': {'Deletion': 3.0, 'Insertion': 8.0, 'SHA': 3},
             '2017-10-19': {'Deletion': 3.0, 'Insertion': 10.0, 'SHA': 19},
             '2017-10-20': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 23},
             '2017-10-21': {'Deletion': 3.0, 'Insertion': 1.0, 'SHA': 44}}}

编辑：@milan 使用的另一个版本：

result = [
    {
         "author": key,
         "commit_activity": group.to_dict(orient="records"),
         "timestamp": [index[1] for index in list(group.index)]
    }
    for key, group in df.groupby("Author")
]

此版本的结果如下所示：

[

    {'author': 'Aaron Pettengill',
      'commit_activity': [{'SHA': 2,
        'Insertion': 156.0,
        'Deletion': 8.0,
        'File path': 2}],
      'timestamp': [Timestamp('2020-05-01 00:00:00+0000', tz='UTC')]},
     {'author': 'Alex Rohleder',
      'commit_activity': [{'SHA': 5,
        'Insertion': 5.0,
        'Deletion': 5.0,
        'File path': 5}],
      'timestamp': [Timestamp('2019-09-06 00:00:00+0000', tz='UTC')]},
     {'author': 'Alex Taylor',
      'commit_activity': [{'SHA': 2,
        'Insertion': 30.0,
        'Deletion': 3.0,
        'File path': 2}],
      'timestamp': [Timestamp('2020-04-29 00:00:00+0000', tz='UTC')]}
]