比较2个DataFrame时出现问题，返回错误结果

Question

There are 2 dfs

df1 and df2

df1 contains:

    account_id  account_name
0   37469426    Name1
1   71508517    Name2
2   85304427    Name3
3   115964688   Name4
4   119853529   Name4

df2 contains:

    account_id  account_name
0   37469426    Name1
1   71508517    Name2
2   85304427    Name3
3   115964688   Name4
4   119853529   Name4
5       1111            Test

I want to compare them, in such way, that in df3 are the values from df1 which are not in df2

In this case it should return nothing.

Datatypes are the same, columns are the same, the number of values differs.

I've tried concat and merge, but the result is wrong.

merged = pd.merge(df1 , df2, on=['account_id', 'account_name'], how='right')

#returns:

    account_id  account_name
0   37469426    Name1
1   71508517    Name2
2   85304427    Name3
3   115964688   Name4
4   119853529   Name5

merged = pd.merge(df1 , df2, on=['account_id', 'account_name'], how='left')

#returns:

0   37469426    Name1
1   71508517    Name2
2   85304427    Name3
3   115964688   Name4
4   119853529   Name4
5       1111            Test

#inner / outer return everything

0   37469426    Name1
1   71508517    Name2
2   85304427    Name3
3   115964688   Name4
4   119853529   Name4
5       1111            Test

compare_ga_accounts = pd.concat([df1 , df2])
compare_ga_accounts.drop_duplicates(keep=False, inplace=True)

#returns:

    account_id  account_name
0   1111            Test

I have no idea why it happens like that((

Answer 1

You can just use isin to compare the column values. For example,

 compare_ga_accounts = df1[~(df1.iloc[:, 0].isin(list(df2.iloc[:, 0])))|(~df1.iloc[:, 1].isin(list(df2.iloc[:, 1])))]