python - 使用 .loc 基于多索引计算更新多索引数据帧中的值
问题描述
我有一个df：
df = pd.DataFrame.from_dict({('group', ''): {0: 'A',
1: 'A',
2: 'A',
3: 'A',
4: 'A',
5: 'A',
6: 'A',
7: 'A',
8: 'A',
9: 'B',
10: 'B',
11: 'B',
12: 'B',
13: 'B',
14: 'B',
15: 'B',
16: 'B',
17: 'B',
18: 'all',
19: 'all'},
('category', ''): {0: 'Amazon',
1: 'Apple',
2: 'Facebook',
3: 'Google',
4: 'Netflix',
5: 'Tesla',
6: 'Total',
7: 'Uber',
8: 'total',
9: 'Amazon',
10: 'Apple',
11: 'Facebook',
12: 'Google',
13: 'Netflix',
14: 'Tesla',
15: 'Total',
16: 'Uber',
17: 'total',
18: 'Total',
19: 'total'},
(pd.Timestamp('2020-06-29 00:00:00'), 'last_sales'): {0: 195.0,
1: 61.0,
2: 106.0,
3: 61.0,
4: 37.0,
5: 13.0,
6: 954.0,
7: 4.0,
8: 477.0,
9: 50.0,
10: 50.0,
11: 75.0,
12: 43.0,
13: 17.0,
14: 14.0,
15: 504.0,
16: 3.0,
17: 252.0,
18: 2916.0,
19: 2916.0},
(pd.Timestamp('2020-06-29 00:00:00'), 'sales'): {0: 1268.85,
1: 18274.385000000002,
2: 19722.65,
3: 55547.255,
4: 15323.800000000001,
5: 1688.6749999999997,
6: 227463.23,
7: 1906.0,
8: 113731.615,
9: 3219.6499999999996,
10: 15852.060000000001,
11: 17743.7,
12: 37795.15,
13: 5918.5,
14: 1708.75,
15: 166349.64,
16: 937.01,
17: 83174.82,
18: 787625.7400000001,
19: 787625.7400000001},
(pd.Timestamp('2020-06-29 00:00:00'), 'difference'): {0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0,
5: 0.0,
6: 0.0,
7: 0.0,
8: 0.0,
9: 0.0,
10: 0.0,
11: 0.0,
12: 0.0,
13: 0.0,
14: 0.0,
15: 0.0,
16: 0.0,
17: 0.0,
18: 0.0,
19: 0.0},
(pd.Timestamp('2020-07-06 00:00:00'), 'last_sales'): {0: 26.0,
1: 39.0,
2: 79.0,
3: 49.0,
4: 10.0,
5: 10.0,
6: 436.0,
7: 5.0,
8: 218.0,
9: 89.0,
10: 34.0,
11: 133.0,
12: 66.0,
13: 21.0,
14: 20.0,
15: 732.0,
16: 3.0,
17: 366.0,
18: 2336.0,
19: 2336.0},
(pd.Timestamp('2020-07-06 00:00:00'), 'sales'): {0: 3978.15,
1: 12138.96,
2: 19084.175,
3: 40033.46000000001,
4: 4280.15,
5: 1495.1,
6: 165548.29,
7: 1764.15,
8: 82774.145,
9: 8314.92,
10: 12776.649999999996,
11: 28048.075,
12: 55104.21000000002,
13: 6962.844999999999,
14: 3053.2000000000003,
15: 231049.11000000002,
16: 1264.655,
17: 115524.55500000001,
18: 793194.8000000002,
19: 793194.8000000002},
(pd.Timestamp('2020-07-06 00:00:00'), 'difference'): {0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0,
5: 0.0,
6: 0.0,
7: 0.0,
8: 0.0,
9: 0.0,
10: 0.0,
11: 0.0,
12: 0.0,
13: 0.0,
14: 0.0,
15: 0.0,
16: 0.0,
17: 0.0,
18: 0.0,
19: 0.0},
(pd.Timestamp('2021-06-28 00:00:00'), 'last_sales'): {0: 96.0,
1: 56.0,
2: 106.0,
3: 44.0,
4: 34.0,
5: 13.0,
6: 716.0,
7: 9.0,
8: 358.0,
9: 101.0,
10: 22.0,
11: 120.0,
12: 40.0,
13: 13.0,
14: 8.0,
15: 610.0,
16: 1.0,
17: 305.0,
18: 2652.0,
19: 2652.0},
(pd.Timestamp('2021-06-28 00:00:00'), 'sales'): {0: 5194.95,
1: 19102.219999999994,
2: 22796.420000000002,
3: 30853.115,
4: 11461.25,
5: 992.6,
6: 188143.41,
7: 3671.15,
8: 94071.705,
9: 6022.299999999998,
10: 7373.6,
11: 33514.0,
12: 35943.45,
13: 4749.000000000001,
14: 902.01,
15: 177707.32,
16: 349.3,
17: 88853.66,
18: 731701.46,
19: 731701.46},
(pd.Timestamp('2021-06-28 00:00:00'), 'difference'): {0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0,
5: 0.0,
6: 0.0,
7: 0.0,
8: 0.0,
9: 0.0,
10: 0.0,
11: 0.0,
12: 0.0,
13: 0.0,
14: 0.0,
15: 0.0,
16: 0.0,
17: 0.0,
18: 0.0,
19: 0.0},
(pd.Timestamp('2021-07-07 00:00:00'), 'last_sales'): {0: 45.0,
1: 47.0,
2: 87.0,
3: 45.0,
4: 13.0,
5: 8.0,
6: 494.0,
7: 2.0,
8: 247.0,
9: 81.0,
10: 36.0,
11: 143.0,
12: 56.0,
13: 9.0,
14: 9.0,
15: 670.0,
16: 1.0,
17: 335.0,
18: 2328.0,
19: 2328.0},
(pd.Timestamp('2021-07-07 00:00:00'), 'sales'): {0: 7556.414999999998,
1: 14985.05,
2: 16790.899999999998,
3: 36202.729999999996,
4: 4024.97,
5: 1034.45,
6: 163960.32999999996,
7: 1385.65,
8: 81980.16499999998,
9: 5600.544999999999,
10: 11209.92,
11: 32832.61,
12: 42137.44500000001,
13: 3885.1499999999996,
14: 1191.5,
15: 194912.34000000003,
16: 599.0,
17: 97456.17000000001,
18: 717745.3400000001,
19: 717745.3400000001},
(pd.Timestamp('2021-07-07 00:00:00'), 'difference'): {0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0,
5: 0.0,
6: 0.0,
7: 0.0,
8: 0.0,
9: 0.0,
10: 0.0,
11: 0.0,
12: 0.0,
13: 0.0,
14: 0.0,
15: 0.0,
16: 0.0,
17: 0.0,
18: 0.0,
19: 0.0}}).set_index(['group','category'])
我正在尝试像这样更新difference索引处的列B, total：
df.loc[('B','total'),(slice(None),'difference')] = df.loc[('B','total'),(slice(None),'last_sales')] /
df.loc[('A','total'),(slice(None),'sales')] * 100
但由于某种原因，我得到了nan值。我不确定为什么会发生这种情况，因为两者
df.loc[('B','total'),(slice(None),'last_sales')]
# and
df.loc[('A','total'),(slice(None),'sales')]
返回值
2020-06-29 00:00:00 last_sales 252.000
2020-07-06 00:00:00 last_sales 366.000
2021-06-28 00:00:00 last_sales 305.000
2021-07-07 00:00:00 last_sales 335.000
2020-06-29 00:00:00 sales 113,731.615
2020-07-06 00:00:00 sales 82,774.145
2021-06-28 00:00:00 sales 94,071.705
2021-07-07 00:00:00 sales 81,980.165
使用df.loc[('B','total'),(slice(None),'last_sales')].values/(df.loc[('A','total'),(slice(None),'sales')] * 100).values我得到：
array([2.21574274e-05, 4.42167056e-05, 3.24220763e-05, 4.08635430e-05])
但是当我使用：
df.loc[('B','total'),(slice(None),'difference')]=df.loc[('B','total'),(slice(None),'last_sales')].to_numpy()/(df.loc[('A','total'),(slice(None),'sales')] * 100).to_numpy()
我得到：
2020-06-29 00:00:00 difference 0.000
2020-07-06 00:00:00 difference 0.000
2021-06-28 00:00:00 difference 0.000
2021-07-07 00:00:00 difference 0.000
解决方案
对于除法，您需要相同MultiIndex的：Series
s1 = df.loc[('B','total'),(slice(None),'last_sales')].rename({'last_sales':'difference'})
s2 = df.loc[('A','total'),(slice(None),'sales')].rename({'sales':'difference'})
df.loc[('B','total'),(slice(None),'difference')] = s1 / s2 * 100
print (df.loc[('B','total'),(slice(None),'difference')])
2020-06-29 00:00:00 difference 0.221574
2020-07-06 00:00:00 difference 0.442167
2021-06-28 00:00:00 difference 0.324221
2021-07-07 00:00:00 difference 0.408635
Name: (B, total), dtype: float64
推荐阅读
- c# - 从 C# 模块调用 C++ 函数会引发随机崩溃
- python - 无法在在线 Jupyter 笔记本中导入 csv 文件
- php - 我在数答案，但我不知道如何自动化
- nginx - 如何让 NGINX 为路径中的特定目录提供服务？
- javascript - 绑定 this - 调用删除项
- javascript - 如何使用公钥和私钥加密条形码上的数据
- python-3.x - Selenium WebDriver 截图不捕获网页页脚
- angular - 插值不适用于点击方法
- image - 如何通过 http post 上传 QPixmap 图像？
- vb.net - VB.net Crystal 报表导出为 html 并使用 Outlook 作为 html 邮件正文发送