在熊猫中合并具有相同行和索引的两个数据帧

Question

I'm trying to merge two pandas dataframes that have common row indexes and common columns 0,1,2 but different column 3, so the resulting dataframe has columns from both:

First dataframe:

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 817 entries, 0 to 816
Data columns (total 3 columns):
0    817 non-null int64
1    817 non-null int64
2    817 non-null float64
dtypes: float64(1), int64(2)
memory usage: 19.2 KB


0   1       2
0   1950    1   -0.060310
1   1950    2   0.626810
2   1950    3   -0.008128
3   1950    4   0.555100
4   1950    5   0.071577

Second dataframe:

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 817 entries, 0 to 816
Data columns (total 3 columns):
0    817 non-null int64
1    817 non-null int64
2    817 non-null float64
dtypes: float64(1), int64(2)
memory usage: 19.2 KB

0   1       2
0   1950    1   0.92
1   1950    2   0.40
2   1950    3   -0.36
3   1950    4   0.73
4   1950    5   -0.59

So far I tried with merge:

pd.merge(df, df2, left_index=True, right_index=True, how='outer')

But results are not what I expect:

    0_x     1_x     2_x     0_y     1_y     2_y
0   1950    1   -0.060310   1950    1   0.92
1   1950    2   0.626810    1950    2   0.40
2   1950    3   -0.008128   1950    3   -0.36
3   1950    4   0.555100    1950    4   0.73
4   1950    5   0.071577    1950    5   -0.59

And with concat:

pd.concat([df, df2], axis=1, ignore_index=True).head()


0   1       2       3       4       5
0   1950    1   -0.060310   1950    1   0.92
1   1950    2   0.626810    1950    2   0.40
2   1950    3   -0.008128   1950    3   -0.36
3   1950    4   0.555100    1950    4   0.73
4   1950    5   0.071577    1950    5   -0.59

I'm expecting something like

0   1       2       3     
0   1950    1   -0.060310    0.92
1   1950    2   0.626810     0.40
2   1950    3   -0.008128    -0.36
3   1950    4   0.555100     0.73
4   1950    5   0.071577     -0.59

EDIT : Maybe I was unclear and I apologize if so, I'm trying to add the last column from the second dataset in the resulting one, so I have the same year, month, value1 and then value2 columns

Answer 1

I would try:

pd.merge(df, df2, on=['0', '1'])

maybe

pd.merge(df, df2, on=[0,1]

Answer 2

Just do:

df.merge(df2, on=1)

you don't need to add index column, once they have same index. And it can be a inner join by default.

Your error was made the merge just by index, the merge function doesn't know that the column 1 is equal in both data.