如何在一个级别上获得前 4 周的销售额

Question

我想找到 Python 中某个水平的前四个星期的销售额。比如说

ID  Category    Date    Sales
1   AA  7/02/2022   1
1   AA  31/01/2022  3
1   AA  24/01/2022  5
1   AA  10/01/2022  7
1   AA  03/01/2022  9
2   BB  7/02/2022   2
2   BB  31/01/2022  4
2   BB  24/01/2022  6
2   BB  17/01/2022  8
2   BB  10/01/2022  10

对于 1 AA 7/02/2022 最后四个星期的总和将为 9（因为 17/01/2022 包不存在并且必须包括当前行日期）

Answer 1

您可以将日期设置为索引，按类别分组并计算 28 天滚动 window 销售额的总和：

import pandas as pd
import io

data = '''ID  Category    Date    Sales
1   AA  7/02/2022   1
1   AA  31/01/2022  3
1   AA  24/01/2022  5
1   AA  10/01/2022  7
1   AA  03/01/2022  9
2   BB  7/02/2022   2
2   BB  31/01/2022  4
2   BB  24/01/2022  6
2   BB  17/01/2022  8
2   BB  10/01/2022  10'''

df = pd.read_csv(io.StringIO(data), sep='\s+')
df['Date'] = pd.to_datetime(df['Date'], format='%d/%m/%Y')

result_df = df.set_index('Date').sort_index().groupby('Category')['Sales'].rolling("28D").sum().reset_index()

输出：

	Category	Date	Sales
0	AA	2022-01-03 00:00:00	9
1	AA	2022-01-10 00:00:00	16
2	AA	2022-01-24 00:00:00	21
3	AA	2022-01-31 00:00:00	15
4	AA	2022-02-07 00:00:00	9
5	BB	2022-01-10 00:00:00	10
6	BB	2022-01-17 00:00:00	18
7	BB	2022-01-24 00:00:00	24
8	BB	2022-01-31 00:00:00	28
9	BB	2022-02-07 00:00:00	20

如何在一个级别上获得前 4 周的销售额

How to get previous 4 week sales at a level

python

pandas

pandas-groupby

pandas-rolling