Question

我想将Count_in列的累积总和添加到按位置，日期和输入时间分组的数据框中。

当前数据框：

我想要的结果：

我尝试了以下操作：

df.groupby(['Location','Date','Entry_Hour']).sum()['Count_in'].groupby(level=1).cumsum().reset_index().tail()

但是结果是错误的：

Answer 1

假设您要使用多索引框架

df = pd.read_csv("cumulative_groupby.csv")
df["Date"] = pd.to_datetime(df["Date"], format="%Y-%m-%d")
df.set_index(["Location", "Date", "Entry_Hour"], inplace=True)
df["cumsum"] = df.groupby(["Location", "Date"]).Count_in.cumsum()
print(df)

输出：

                                Count_out  Count_in  cumsum
Location Date       Entry_Hour                             
YEMEN    2018-10-29 16                300       500     500
                    17                200       600    1100
                    18                 10        20    1120
         2018-10-30 16                400        20      20
                    17                500        20      40
                    18                700        20      60
USA      2018-10-29 2                 300       500     500
                    3                 200       600    1100
                    4                  10       456    1556
         2018-10-30 2                 400       123     123
                    3                 500         6     129
                    4                 700       788     917

cumulative_groupby.csv

Date,Entry_Hour,Count_out,Location,Count_in
2018-10-29,16,300,YEMEN,500
2018-10-29,17,200,YEMEN,600
2018-10-29,18,10,YEMEN,20
2018-10-30,16,400,YEMEN,20
2018-10-30,17,500,YEMEN,20
2018-10-30,18,700,YEMEN,20
2018-10-29,2,300,USA,500
2018-10-29,3,200,USA,600
2018-10-29,4,10,USA,456
2018-10-30,2,400,USA,123
2018-10-30,3,500,USA,6
2018-10-30,4,700,USA,788

熊猫groupby累积

1 个答案: