如何使用python-pandas合并基于互信息的两个数据框？

Question

Given two data frame, df1 and df2 , containing information of item_id-rating and item_id-class :

df1:

B0006IYIMW 5.0
B000A56PUO 3.0
B000AMLQQU 4.0
B000OVNMGE 1.0

df2:

B0006IYIMW iphone
B000OVNMGE samsung
B000AMLQQU htc
B000A56PUO nokia

I wish to merge df1 and df to get the full info of item_id-class-rating , so the resulting data frame should be:

B0006IYIMW iphone 5.0
B000OVNMGE samsung 1.0
B000AMLQQU htc 4.0
B000A56PUO nokia 3.0

Please notice that the order of two data frames maybe different.

Could you please tell me how to do it? Thanks in advance!

Answer 1

Try this:

import pandas as pd

df1 = pd.DataFrame([['B0006IYIMW',5.0],['B000A56PUO', 3.0],['B000AMLQQU', 4.0],['B000OVNMGE', 1.0]],columns=('item_id','rating'))
df2 = pd.DataFrame([['B0006IYIMW','iphone'],['B000A56PUO', 'nokia'],['B000AMLQQU', 'htc'],['B000OVNMGE', 'samsung']],columns=('item_id','class'))

df_merged = df1.merge(df2,on='item_id')

print df_merged

Answer 2

As usually, when I can't find the solution I start to hack my own, and by the time I've achieved many bad results and finally have the right one, somebody else already posted one-line solution :) Here it is anyway

import pandas as pd
# the frames are named the same way, and rows are in the same order
# assuming item-ids are unique I've created list of indices
# which corresponds to the index of the elements from df1 in df2
df2_index = [df2['item-id'].tolist().index(df1['item-id'][x]) for x in range(len(df1))]
# now reindex df1 according to the list and reset index!
df1 = df1.reindex(df2_index).reset_index(drop=True)
# now you can simply add the missing column
df2['item-rating'] = df1['item-rating']

Answer 3

Setup

import pandas as pd

idx = pd.Index(['B0006IYIMW', 'B000A56PUO', 'B000AMLQQU', 'B000OVNMGE'],
               name='item-id')
df1 = pd.DataFrame([5., 3., 4., 1.],
                   columns=['rating'], index=idx)
df2 = pd.DataFrame(['iphone', 'samsung', 'htc', 'nokia'],
                   columns=['class'], index=idx)

Solution

df = pd.concat([df2, df1], axis=1)

Demonstration

print df 

              class  rating
item-id                    
B0006IYIMW   iphone     5.0
B000A56PUO  samsung     3.0
B000AMLQQU      htc     4.0
B000OVNMGE    nokia     1.0