在具有行规范的熊猫中运行总和答案

【问题标题】：Running sums in pandas with row specification在具有行规范的熊猫中运行总和
【发布时间】：2022-01-20 19:11:32
【问题描述】：

我有一些数据，我试图计算每个批次的每个批次项目编号的所有计数的总测量值和计数 2、3 和 4 的测量值总和。理想情况下，我会在原始数据上有 2 个额外的列，其中包含总测量值和计数 2、3 和 4 的测量值——即使这些值会重复，它们也会用每条记录表示。以下是数据集的示例：

Date    Sample Type Lot #   Lot item #  Count   Measurement
0   2021-12-05  G   ABS123-G    1   1   5.0
1   2021-12-05  G   ABS123-G    1   2   3.0
2   2021-12-05  G   ABS123-G    1   3   7.0
3   2021-12-05  G   ABS123-G    1   4   25.1
4   2021-12-05  G   ABS123-G    1   5   66.0
5   2021-12-05  G   ABS123-G    1   6   54.0
6   2021-12-05  G   ABS123-G    1   7   12.0
7   2021-12-05  G   ABS123-G    1   8   0.0
8   2021-12-05  G   ABS123-G    1   9   1.0
9   2021-12-05  G   ABS123-G    1   10  5.0
10  2021-12-05  G   ABS123-G    2   1   2.0
11  2021-12-05  G   ABS123-G    2   2   4.0
12  2021-12-05  G   ABS123-G    2   3   889.0
13  2021-12-05  G   ABS123-G    2   4   12.4
14  2021-12-05  G   ABS123-G    2   5   51.4
15  2021-12-05  G   ABS123-G    2   6   12.0
16  2021-12-05  G   ABS123-G    2   7   14.0
17  2021-12-05  G   ABS123-G    2   8   2.0
18  2021-12-05  G   ABS123-G    2   9   1.0
19  2021-12-05  G   ABS123-G    2   10  0.1
20  2021-12-05  B   ABS123-B    1   1   4.0
21  2021-12-05  B   ABS123-B    1   2   58.0
22  2021-12-05  B   ABS123-B    1   3   123.0
23  2021-12-05  B   ABS123-B    1   4   12.5
24  2021-12-05  B   ABS123-B    1   5   11.0
25  2021-12-05  B   ABS123-B    1   6   135.5
26  2021-12-05  B   ABS123-B    1   7   17.0
27  2021-12-05  B   ABS123-B    1   8   1.0
28  2021-12-05  B   ABS123-B    1   9   5.0
29  2021-12-05  B   ABS123-B    1   10  0.3

我的方法是尝试将计数过滤为 2,3,4，计算总和，然后根据批次和批次项目 # 将 df 加入原始数据，然后对总数执行类似的操作。但是，我在尝试求和时遇到了错误。

df2 = df.loc[(df['Count'] == 2) | (df['Count'] == 3) | (df['Count'] == 4)]
df2['Counts 2,3,4'] = df2.grouby(['Lot #, 'Lot item #'])['Measurement'].sum()
df2

TypeError: incompatible index of inserted column with frame index

过滤器有效，但第二部分无效。首先，我不知道是什么导致了错误，我需要重置索引吗？另外，这是正确的方法吗？欢迎提出任何建议。

【问题讨论】：

能否展示一个包含预期输出的示例数据框？

标签： python pandas sum typeerror

【解决方案1】：

我们可以使用isin 通过定义整数值列表来简化相等性检查。然后我们可以在groupby sum 之后使用join 并指定要加入的列。最后renameSeries 为新的列名：

# Specify columns to groupby and join back on
grp_cols = ['Lot #', 'Lot item #']
joined = df.join(
    df[df['Count'].isin([2, 3, 4])]  # Values to include
        .groupby(grp_cols)['Measurement'].sum()  # Take sum per group
        .rename('Counts 2,3,4'),  # Specify new column name
    on=grp_cols,
)

joined:

          Date Sample Type     Lot #  Lot item #  Count  Measurement  Counts 2,3,4
0   2021-12-05           G  ABS123-G           1      1          5.0          35.1
1   2021-12-05           G  ABS123-G           1      2          3.0          35.1
2   2021-12-05           G  ABS123-G           1      3          7.0          35.1
3   2021-12-05           G  ABS123-G           1      4         25.1          35.1
4   2021-12-05           G  ABS123-G           1      5         66.0          35.1
5   2021-12-05           G  ABS123-G           1      6         54.0          35.1
6   2021-12-05           G  ABS123-G           1      7         12.0          35.1
7   2021-12-05           G  ABS123-G           1      8          0.0          35.1
8   2021-12-05           G  ABS123-G           1      9          1.0          35.1
9   2021-12-05           G  ABS123-G           1     10          5.0          35.1
10  2021-12-05           G  ABS123-G           2      1          2.0         905.4
11  2021-12-05           G  ABS123-G           2      2          4.0         905.4
12  2021-12-05           G  ABS123-G           2      3        889.0         905.4
13  2021-12-05           G  ABS123-G           2      4         12.4         905.4
14  2021-12-05           G  ABS123-G           2      5         51.4         905.4
15  2021-12-05           G  ABS123-G           2      6         12.0         905.4
16  2021-12-05           G  ABS123-G           2      7         14.0         905.4
17  2021-12-05           G  ABS123-G           2      8          2.0         905.4
18  2021-12-05           G  ABS123-G           2      9          1.0         905.4
19  2021-12-05           G  ABS123-G           2     10          0.1         905.4
20  2021-12-05           B  ABS123-B           1      1          4.0         193.5
21  2021-12-05           B  ABS123-B           1      2         58.0         193.5
22  2021-12-05           B  ABS123-B           1      3        123.0         193.5
23  2021-12-05           B  ABS123-B           1      4         12.5         193.5
24  2021-12-05           B  ABS123-B           1      5         11.0         193.5
25  2021-12-05           B  ABS123-B           1      6        135.5         193.5
26  2021-12-05           B  ABS123-B           1      7         17.0         193.5
27  2021-12-05           B  ABS123-B           1      8          1.0         193.5
28  2021-12-05           B  ABS123-B           1      9          5.0         193.5
29  2021-12-05           B  ABS123-B           1     10          0.3         193.5

示例 DataFrame 构造函数：

import pandas as pd

df = pd.DataFrame({
    'Date': pd.to_datetime(
        ['2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05',
         '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05',
         '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05',
         '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05',
         '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05',
         '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05', '2021-12-05']),
    'Sample Type': ['G', 'G', 'G', 'G', 'G', 'G', 'G', 'G', 'G', 'G', 'G', 'G',
                    'G', 'G', 'G', 'G', 'G', 'G', 'G', 'G', 'B', 'B', 'B', 'B',
                    'B', 'B', 'B', 'B', 'B', 'B'],
    'Lot #': ['ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G',
              'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G',
              'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G',
              'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G', 'ABS123-G',
              'ABS123-B', 'ABS123-B', 'ABS123-B', 'ABS123-B', 'ABS123-B',
              'ABS123-B', 'ABS123-B', 'ABS123-B', 'ABS123-B', 'ABS123-B'],
    'Lot item #': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
                   1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
    'Count': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1,
              2, 3, 4, 5, 6, 7, 8, 9, 10],
    'Measurement': [5.0, 3.0, 7.0, 25.1, 66.0, 54.0, 12.0, 0.0, 1.0, 5.0, 2.0,
                    4.0, 889.0, 12.4, 51.4, 12.0, 14.0, 2.0, 1.0, 0.1, 4.0,
                    58.0, 123.0, 12.5, 11.0, 135.5, 17.0, 1.0, 5.0, 0.3]
})

【讨论】：

【解决方案2】：

我能够弄清楚是因为这是索引的问题。当我刚刚删除新列名并运行 groupby 时，它起作用了。然后我将索引重置为 groupby 并且我能够毫无问题地合并到原始 df 。与总数相同。

df2 = df.loc[(df['Count'] == 2) | (df['Count'] == 3) | (df['Count'] == 4)]
df3 = df2.groupby(['Lot #', 'Lot item #'])['Measurement'].sum()
df3 = df3.reset_index()

joined = pd.merge(df, df3, how='left', left_on=['Lot #', 'Lot item #'], right_on=['Lot #', 'Lot item #'])

我只是认为肯定有比这更优雅的解决方案吗？但也许不是？

【讨论】：