pandas中比较两个dataframe并输出差异部分？

[MyGaoRen]

有两个文件，格式一样，每行有两列，一列标签，一列对应数字。两边行数不同，两边标签大部分相同。需要找出两边差异，输出到一个新文件。

使用OS命令先sort ，在diff -u 1.txt 2.txt可以合并一个新文件，再基于"+""-"关键字grep过滤。

用python中用 pandas 对比，转化两个dataframe对比并输出差异部分，结果总是不符合要求。

 

文件1.txt

e39aa0582fb3d25b42a77fb9cb09472d 99
e39aa0582fb3d25b42a77fb9cb094721 99
e39aa0582fb3d25b42a77fb9cb094722 98
e39aa0582fb3d25b42a77fb9cbc94722 99
e39aa0582fb3d25b42a77fb9cb094723 99
e39aa0582fb3d25b42a77fb9cb094724 99
e39aa0582fb3d25b42a77fb9cb094725 99

文件2.txt

e39aa0582fb3d25b42a77fb9cb09472d 99
e39aa0582fb3d25b42a77fb9cb094721 99
e39aa0582fb3d25b42a77fb9cb094722 99
e39aa0582fb3d25b42a77fb9cb094723 99
e39aa0582fb3d25b42a77fb9cb094724 98
e39aa0582fb3d25b42a77fb9cb094734 98
e39aa0582fb3d25b42a77fb9cb094725 99

希望能对比两个文件不同，并输出差异，希望效果如下，按照uuid关联两个文件，仅显示有差异的

                           uuid1    num1  
e39aa0582fb3d25b42a77fb9cb094722    98.0
e39aa0582fb3d25b42a77fb9cbc94722    99.0
e39aa0582fb3d25b42a77fb9cb094724    99.0
e39aa0582fb3d25b42a77fb9cb094734     NaN

结果最好类似如下

                           uuid1  num1_x  num1_y
e39aa0582fb3d25b42a77fb9cb094722    98.0    99.0
e39aa0582fb3d25b42a77fb9cbc94722    99.0     NaN
e39aa0582fb3d25b42a77fb9cb094724    99.0    98.0
e39aa0582fb3d25b42a77fb9cb094734     NaN    98.0

使用pandas dataframe的compare,diff无法达到效果，以下是代码样例

import pandas as pd

fd1=pd.read_table('1.txt', sep=' ',header=None)
fd1.columns=['uuid1','num1']
fd2=pd.read_table('2.txt', sep=' ',header=None)
fd2.columns=['uuid1','num1']


merge1 = pd.merge(fd1,fd2,how='outer',on='uuid1')

compare1=fd2.compare(fd1, align_axis=1, keep_shape=True, keep_equal=True)

merge1['diff']=merge1['num1_x']-merge1['num1_y']

 

[cryincold]

顶