通过将另一个数据框与一对多关系进行匹配来创建新列的数据框

Question

I am pretty new using pandas library and I am not used to dataframe yet. I am trying to add a column to a dataframe1 by using a value of a column from dataframe1, use this value as an index of dataframe2, and get the corresponding value

I have two dataframes:

df1 = pandas.DataFrame({'customer' : pd.Series([28, 28, 29, 30],
                                     index=['0', '1', '3', '4']),
                        'store' : pd.Series([14, 14, 14, 22], 
                                  index=['0', '1', '3', '4'])})

df2 = pandas.DataFrame({'value': pd.Series([6, 7, 8], 
                                 index=[0, 1, 2]),
                         'store': pd.Series([14, 14, 22], 
                                  index=[0,1, 2])})

df2.groupby(['store']).agg({'Value':[sum]})

My goal is to add in df1 a column containing the 'values' in the index corresponding to the value df2 when 'store' values of df1

Expecting output:

df3 = {'customer' : pd.Series([28., 28., 29., 30.], index=['0', '1', '3', '4']),
'store' : pd.Series([14, 14, 14, 22], index=['0', '1', '3', '4']),
'value' : pd.Series([6, 6, 6, 8], index=['0', '1', '3', '4']}

I tried:

for index, row in df1.iterrows():
    df1['Values'] = df2.loc[row['store']]

But I get the TypeError: incompatible index of inserted column with frame index

for index, row in df1.iterrows():
    df1['Values'] = df2.loc[pd.Index(row['store'])]

But I get a TypeError:

Index(...) must be called with a collection of some kind, 'int' was passed

Thank you very much for your help, I am really struggling on that

Answer 1

Let's change your groupby statement to create a pd.Series and use map :

s = df2.groupby(['store'])['value'].agg('sum')
df1['value'] = df1['store'].map(s)

df1 

Output:

   customer  store  value
0        28     14     13
1        28     14     13
3        29     14     13
4        30     22      8

Answer 2

在将df2汇总为唯一store值之后，这对我有用：

df1['value'] = [int(df2[df2.store==s].value) for s in df1.store]

Answer 3

You simply need:

df1.merge(df2.reset_index(), how='left', on=['store'])

Output:

  customer  store   value
0   28       14      13
1   28       14      13
2   29       14      13
3   30       22       8