合并并组合两列不同的数据帧

Question

I have 2 dataframes :

ID             word
1              srv1
2              srv2
3              srv1
4              nan
5              srv3
6              srv1
7              srv5
8              nan

ID             word
1              nan
2              srv12
3              srv10
4              srv8
5              srv4
6              srv7
7              nan
8              srv9

What I need is to merge thoses 2 dataframes on ID and combine the column word to get :

ID             word
1              srv1 
2              srv2 , srv12
3              srv1 , srv10
4              srv8
5              srv3 , srv4
6              srv1 , srv7
7              srv5
8              srv9

With the following code

merge = pandas.merge(df1,df2,on="ID",how="left")
merge["word"] = merge[word_x] + " , " + merge["word_y"]

I am getting:

ID             word
1              nan 
2              srv2 , srv12
3              srv1 , srv10
4              nan
5              srv3 , srv4
6              srv1 , srv7
7              nan
8              nan

Which it is not the correct solution.

Answer 1

You can use Series.str.cat and the na_rep option to populate the word column even if one of the source columns in nan , then use str.strip to trim any leading/trailing ' , ' not between words.

m['word'] = m['word_x'].str.cat(m['word_y'], sep=' , ', na_rep='').str.strip(' , ')

returns

   ID word_x word_y          word
0   1   srv1    NaN          srv1
1   2   srv2  srv12  srv2 , srv12
2   3   srv1  srv10  srv1 , srv10
3   4    NaN   srv8          srv8
4   5   srv3   srv4   srv3 , srv4
5   6   srv1   srv7   srv1 , srv7
6   7   srv5    NaN          srv5
7   8    NaN   srv9          srv9

Answer 2

you can use np.select to select the existing value, or the concatenated value.

try this:

import pandas as pd
import numpy as np
from io import StringIO

df1 = pd.read_csv(StringIO("""
ID             word
1              srv1
2              srv2
3              srv1
4              nan
5              srv3
6              srv1
7              srv5
8              nan"""), sep=r"\s+")

df2 = pd.read_csv(StringIO("""
ID             word
1              nan
2              srv12
3              srv10
4              srv8
5              srv4
6              srv7
7              nan
8              srv9"""), sep=r"\s+")


conditions = [(~df1["word"].isna()) & df2["word"].isna(), df1["word"].isna() & (~df2["word"].isna()), (~df1["word"].isna()) & (~df2["word"].isna())]
choices = [df1["word"], df2["word"], df1["word"] + "," + df2["word"]]

df1["word"] = np.select(conditions,choices)

print(df1)

Output:

   ID        word
0   1        srv1
1   2  srv2,srv12
2   3  srv1,srv10
3   4        srv8
4   5   srv3,srv4
5   6   srv1,srv7
6   7        srv5
7   8        srv9

Answer 3

Based on what I think you want to do I would first get rid of those nan 's:

df_1.fillna(value="")
df_2.fillna(value="")

And then I would try the merge again and see if you get what you want.