尝试 cumsum() pandas 具有相同值的数据框出现在多列中

Question

我正在尝试使用 groupby 获取累计和，其中累计和应用于包含相同值的多个列

import pandas as pd
import numpy as np

df = pd.DataFrame([['Jazz', 'Clippers', 89, 100],
                              ['Clippers' , 'Jazz', 101, 97],
                              ['Bucks' , 'Jazz', 99, 112],
                              ['Jazz' , 'Bucks', 109, 88]],
                      columns=['home_team', 'away_team', 'home_points', 'away_points'])
print(df)

这将生成一个输出为

的数据帧

  home_team away_team  home_points  away_points
0      Jazz  Clippers           89          100
1  Clippers      Jazz          101           97
2     Bucks      Jazz           99          112
3      Jazz     Bucks          109           88

我想做的是获得主队和客队的累计总分，这将说明每支球队同时出现在主队和客队列中的事实，但我所能弄清楚的是按球队名称分组的累计总数，每个球队的主场或客场总和，如下所示

df["home_cumulative_points"]= df.groupby(["home_team"])["home_points"].cumsum() 
df["away_cumulative_points"]= df.groupby(["away_team"])["away_points"].cumsum() 
print(df)

产生

  home_team away_team  home_points  away_points  home_cumulative_points  away_cumulative_points
0      Jazz  Clippers           89          100                      89                     100
1  Clippers      Jazz          101           97                     101                      97
2     Bucks      Jazz           99          112                      99                     209
3      Jazz     Bucks          109           88                     198                      88

有什么方法可以让我通过 groupby 计算主场和客场列中同一支球队的累计总和，从而使运行总和加上球队的积分，而不管他们是否在主场还是离开？所以最后一行的理想输出是

  home_team away_team  home_points  away_points  home_cumulative_points  away_cumulative_points
3      Jazz     Bucks          109           88                     407                      187

我猜我可能需要做一个 for 循环或其他什么，但我不确定如何最好地去做。提前感谢您的任何反馈！

Answer 1

想法是 select 唯一必要的列，按 _ 拆分为 MultiIndex，按 DataFrame.stack 重塑，因此可以对每个列一起使用 cumsum :

cols = ['home_team', 'away_team', 'home_points', 'away_points']

df1 = df[cols].copy()
df1.columns = df1.columns.str.split('_', expand=True)
df1 = df1.stack(0).rename_axis(['lev1','lev2'])
df1["cumulative_points"]= df1.groupby(["team", 'lev1'])["points"].cumsum() 

df2 = df1.unstack()
df2.columns = df2.columns.map(lambda x: f'{x[1]}_{x[0]}')
print(df2)
      away_points  home_points away_team home_team  away_cumulative_points  \
lev1                                                                         
0             100           89  Clippers      Jazz                     100   
1              97          101      Jazz  Clippers                      97   
2             112           99      Jazz     Bucks                     112   
3              88          109     Bucks      Jazz                      88   

      home_cumulative_points  
lev1                          
0                         89  
1                        101  
2                         99  
3                        109

或：

df["home_cumulative_points"]= df1.loc['home', 'cumulative_points']
df["away_cumulative_points"]= df1.loc['away', 'cumulative_points']

另一种方法是使用 concat 和 rename 进行整形：

f = lambda x: x.split('_')[1]
df1 = pd.concat([df[['home_team', 'home_points']].rename(columns=f),
                 df[['away_team', 'away_points']].rename(columns=f)], keys=('home','away'))
df1 = df1.rename_axis(['lev1','lev2'])
df1["cumulative_points"]= df1.groupby(["team", 'lev1'])["points"].cumsum()

df["home_cumulative_points"]= df1.loc['home', 'cumulative_points']
df["away_cumulative_points"]= df1.loc['away', 'cumulative_points']
print(df)
  home_team away_team  home_points  away_points  home_cumulative_points  \
0      Jazz  Clippers           89          100                      89   
1  Clippers      Jazz          101           97                     101   
2     Bucks      Jazz           99          112                      99   
3      Jazz     Bucks          109           88                     198   

   away_cumulative_points  
0                     100  
1                      97  
2                     209  
3                      88

尝试 cumsum() pandas 具有相同值的数据框出现在多列中

Trying to cumsum() pandas dataframe with same values appearing in multiple columns

python

dataframe

pandas

cumsum

pandas-groupby