How can I merge together several pandas dataframes on a certain column without 'pandas.merge'?

问题

I often find myself with several pandas dataframes in the following form:

import pandas as pd
df1 = pd.read_table('filename1.dat')
df2 = pd.read_table('filename2.dat')
df3 = pd.read_table('filename3.dat')

print(df1)
columnA   first_values
name1     342
name2     822
name3     121
name4     3434

print(df2)
columnA   second_values
name1     8
name2     1
name3     1
name4     2

print(df3)
columnA   third_values
name1     910
name2     301
name3     132
name4     299

I would like to merge together each of these dataframes on 'columnA', giving

columnA   first_values    second_values    third_values
name1     342             8                910
name2     822             1                301
name3     121             1                132
name4     3434            2                299

I normally resort to this hack:

merged1 = df1.merge(df2, on='columnA')

then

merged2 = df3.merge(merged1, on='columnA')

But this doesn't scale for many dataframes. What is the correct way to do this?

回答1:

You can set columnA as the index and concat (reset index at the end):

dfs = [df1, df2, df3]

pd.concat([df.set_index('columnA') for df in dfs], axis=1).reset_index()
Out: 
  columnA  first_values  second_values  third_values
0   name1           342              8           910
1   name2           822              1           301
2   name3           121              1           132
3   name4          3434              2           299

回答2:

Assuming that the three dataframes have the same index, you could just add columns to get the desired dataframes and not worry about merging, like so,

import pandas as pd

#create the dataframe
colA = ['name1', 'name2', 'name3', 'name4']
first = [ 342, 822, 121, 3434]
second = [ 8,1,1,2]
third = [ 910,301,132, 299]
df1 = pd.DataFrame({'colA': colA, 'first': first})
df2 = pd.DataFrame({'colA': colA, 'second': second})
df3 = pd.DataFrame({'colA': colA, 'third': third})


df_merged = df1.copy()
df_merged['second']= df2.second
df_merged['third']= df3.third
print (df_merged.head())

    colA  first  second  third
0  name1    342       8    910
1  name2    822       1    301
2  name3    121       1    132
3  name4   3434       2    299

来源：https://stackoverflow.com/questions/38775588/how-can-i-merge-together-several-pandas-dataframes-on-a-certain-column-without