使用自定义函数将年数据分解为月数据

Question

我正在尝试按费用将年度订阅细分为月度订阅。

示例数据集-

import numpy as np
import pandas as pd

df = pd.DataFrame({
    'Customer_ID': [1, 2, 3, 4, 5],
    'Plan' : ['Yearly', 'Monthly', 'Monthly', 'Yearly', 'Yearly'],
    'Join_Date': ['1/10/2020', '1/15/2020', '2/21/2020', '2/21/2020', '3/09/2020'],
    'Fee' : [120, 12, 18, 86, 144]
})

df['Join_Date'] = pd.to_datetime(df['Join_Date'])

df

在这里，客户 1 在 2020 年 1 月到 2021 年 1 月之间的年度订阅费为 120 美元。我希望我的数据框将 2020-01 和 2020-12 之间的费用细分为 10 美元（120 美元/12 个月）显示当年每个月的月费 ($10)。

我试了一堆重采样的方法，都没有用。一个例子-

def atom(row):
    if df.Plan=='Yearly':
        return (df.Fee/12)

df.groupby(pd.Grouper(key='Join_Date', freq='1M')).apply(atom)

第一个客户的预期输出-

还有别的方法吗？

Answer 1

您在找这样的东西吗？

import pandas as pd
df = pd.DataFrame({
    'Cutomer_ID': [1, 2, 3, 4, 5],
    'Plan' : ['Yearly', 'Monthly', 'Monthly', 'Yearly', 'Yearly'],
    'Join_Date': ['1/10/2020', '1/15/2020', '2/21/2020', '2/21/2020', '3/09/2020'],
    'Fee' : [120, 12, 18, 86, 144]
})

df['Join_Date'] = pd.to_datetime(df['Join_Date'])

df['Monthly_Fee'] = df['Fee']
df.loc[df['Plan'] == 'Yearly','Monthly_Fee'] = (df.Fee/12).round(2)

print (df)

这个输出将是：

   Cutomer_ID     Plan  Join_Date  Fee  Monthly_Fee
0           1   Yearly 2020-01-10  120        10.00
1           2  Monthly 2020-01-15   12        12.00
2           3  Monthly 2020-02-21   18        18.00
3           4   Yearly 2020-02-21   86         7.17
4           5   Yearly 2020-03-09  144        12.00

Answer 2

首先将年度记录扩大np.repeat()。然后在 df1["Plan"] == "Yearly" 上有选择地执行以下操作：

月费可直接计算
月份的增量可以使用groupby-cumcount获得并映射到. Such method receives a PerformanceWarning, which can be suppressed （代码中省略）。

代码

# expand the Yearly records
df1 = df.loc[np.repeat(df.index, df["Plan"].map({"Yearly": 12, "Monthly":1}))]

# compute monthly fee and join date 
df1.loc[df1["Plan"] == "Yearly", "Fee"] /= 12
df1.loc[df1["Plan"] == "Yearly", "Join_Date"] += \
    df1.groupby(["Customer_ID", "Plan"]).cumcount()\
       .loc[df1["Plan"] == "Yearly"]\
       .map(lambda i: pd.DateOffset(months=i))

结果

print(df1)
   Customer_ID     Plan  Join_Date        Fee
0            1   Yearly 2020-01-10  10.000000
0            1   Yearly 2020-02-10  10.000000
0            1   Yearly 2020-03-10  10.000000
0            1   Yearly 2020-04-10  10.000000
0            1   Yearly 2020-05-10  10.000000
0            1   Yearly 2020-06-10  10.000000
0            1   Yearly 2020-07-10  10.000000
0            1   Yearly 2020-08-10  10.000000
0            1   Yearly 2020-09-10  10.000000
0            1   Yearly 2020-10-10  10.000000
0            1   Yearly 2020-11-10  10.000000
0            1   Yearly 2020-12-10  10.000000
1            2  Monthly 2020-01-15  12.000000
2            3  Monthly 2020-02-21  18.000000
3            4   Yearly 2020-02-21   7.166667
3            4   Yearly 2020-03-21   7.166667
3            4   Yearly 2020-04-21   7.166667
3            4   Yearly 2020-05-21   7.166667
3            4   Yearly 2020-06-21   7.166667
3            4   Yearly 2020-07-21   7.166667
3            4   Yearly 2020-08-21   7.166667
3            4   Yearly 2020-09-21   7.166667
3            4   Yearly 2020-10-21   7.166667
3            4   Yearly 2020-11-21   7.166667
3            4   Yearly 2020-12-21   7.166667
3            4   Yearly 2021-01-21   7.166667
4            5   Yearly 2020-03-09  12.000000
4            5   Yearly 2020-04-09  12.000000
4            5   Yearly 2020-05-09  12.000000
4            5   Yearly 2020-06-09  12.000000
4            5   Yearly 2020-07-09  12.000000
4            5   Yearly 2020-08-09  12.000000
4            5   Yearly 2020-09-09  12.000000
4            5   Yearly 2020-10-09  12.000000
4            5   Yearly 2020-11-09  12.000000
4            5   Yearly 2020-12-09  12.000000
4            5   Yearly 2021-01-09  12.000000
4            5   Yearly 2021-02-09  12.000000

使用自定义函数将年数据分解为月数据

Breaking down year to monthly data with custom function

python

pandas

pandas-groupby

resampling

代码

结果