合并两个具有不同形状的数据框中的列

Question

I have two df such as these:

  dfA
Out[191]: 
   a  b  c  d
0  N  M  1  3
1  S  F  2  4
1  S  F  2  4

And another one like this:

dfM
Out[192]: 
   X  Y   d1   d2   d3
0  N  M  0.1  0.2  0.3
1  S  F  1.0  2.0  3.0

Now I want to merge these two to get a df like this:

   a  b  c  d  e
0  N  M  1  3  0.1
1  S  F  2  4  1.0
1  S  F  2  4  2.0

The merged df has value from dfM d's columns which got filled based on number of times the rows of dfA got repeated. How to do this in python?

Answer 1

One possible solution is, for each dfM row, use their X and Y values to filter dfA rows and set 'e' column to the remaining values of dfM rows list. Check the example below:

for i, row in dfM.iterrows():
    d_values = row[2:].tolist()
    indexes = list(dfA[(dfA.a == row.X) & (dfA.b == row.Y)].index)
    dfA.loc[indexes, "e"] = d_values[:len(indexes)]

Answer 2

You can use cumcount for helper counter column for merge with left join and also second DataFrame is reshaped by melt :

dfA['groups'] = dfA.groupby(['a','b']).cumcount()

dfM1 = dfM.melt(['X','Y'], value_name='e')
dfM1['groups'] = dfM1.groupby(['X','Y']).cumcount()
print (dfM1)
   X  Y variable    e  groups
0  N  M       d1  0.1       0
1  S  F       d1  1.0       0
2  N  M       d2  0.2       1
3  S  F       d2  2.0       1
4  N  M       d3  0.3       2
5  S  F       d3  3.0       2

d = {'X':'a', 'Y':'b'}
df = (dfA.merge(dfM1.rename(columns=d), on=['a','b', 'groups'], how='left')
         .drop(['variable','groups'],axis=1))
print (df)
   a  b  c  d    e
0  N  M  1  3  0.1
1  S  F  2  4  1.0
2  S  F  2  4  2.0