熊猫数据框按两列分组,计数和总和

时间:2020-08-10 16:39:58

标签: python pandas dataframe

我有以下df,我想按“名称”分组,所以有一个“ A”和“ B”计数列以及一个“总销售额”总和列:

例如将其旋转:

data = {'A or B' : ['A','A','B','B','A','B'],
        'Name' : ['Ben','Ben','Ben','Sam','Sam','Sam'],
        'Sales ($)' : [10,5,2,5,6,7]
       }

df=pd.DataFrame(data, columns = ['A or B','Name','Sales ($)'])

所以看起来像这样:

grouped_data = {'A' : [2,1],
        'B' : [1,2],
        'Name' : ['Ben','Sam'],
        'Total Sales ($)' : [17,18]
       }

df=pd.DataFrame(grouped_data, columns = ['A','B','Name','Total Sales ($)'])

3 个答案:

答案 0 :(得分:3)

您可以尝试使用pd.get_dummiesjoingroupby + sum

pd.get_dummies(df['A or B'])\
  .join(df.drop('A or B',1))\
  .groupby('Name',as_index=False).sum()

输出:

  Name  A  B  Sales ($)
0  Ben  2  1         17
1  Sam  1  2         18

详细信息:

首先,使用get_dummies将分类变量转换为虚拟变量/指标变量:

pd.get_dummies(df['A or B'])
#   A  B
#0  1  0
#1  1  0
#2  0  1
#3  0  1
#4  1  0
#5  0  1

然后使用join,在'A or B'列被删除的情况下,使用原始df来连接假人:

pd.get_dummies(df['A or B']).join(df.drop('A or B',1))
#   A  B Name  Sales ($)
#0  1  0  Ben         10
#1  1  0  Ben          5
#2  0  1  Ben          2
#3  0  1  Sam          5
#4  1  0  Sam          6
#5  0  1  Sam          7

最后,根据名称执行groupby + sum

pd.get_dummies(df['A or B']).join(df.drop('A or B',1)).groupby('Name',as_index=False).sum()
#  Name  A  B  Sales ($)
#0  Ben  2  1         17
#1  Sam  1  2         18

答案 1 :(得分:1)

您可以在import pandas as pd import matplotlib.pyplot as plt data = [[1, 10, 'red'], [2, 15, 'green'], [3, 14, 'blue']] df = pd.DataFrame(data, columns = ['x', 'y', 'color']) fig, ax = plt.subplots() for i in df.index: ''' Get two rows each time, every row has a point (x, y) Two points can draw a line, use the color defined by first row ''' partial = df.iloc[i:i+2, :] ax.plot(partial['x'], partial['y'], color=partial['color'].iloc[0], zorder = 0) plt.show() 内使用聚合

groupby

答案 2 :(得分:0)

分步解决方案:

import pandas as pd
data = {'A or B' : ['A','A','B','B','A','B'],
        'Name' : ['Ben','Ben','Ben','Sam','Sam','Sam'],
        'Sales ($)' : [10,5,2,5,6,7]
       }

df=pd.DataFrame(data, columns = ['A or B','Name','Sales ($)'])

#first create dummy for 'A or B' column
y = pd.get_dummies(df['A or B'])

#concatenate with original data frame
df=pd.concat([y,df], axis=1)
#delete the column
del df['A or B']

#now do the group by
df=df.groupby('Name').agg({'A':'sum',
                         'B':'sum', 
                         'Sales ($)': 'sum'})

#reset the index
df.reset_index(level=0, inplace=True)
print(df)

输出:

  Name  A  B  Sales ($)
0  Ben  2  1         17
1  Sam  1  2         18