I have 2 dataframe and I want to join them on the basis of month and year from a date without creating extra columns:

example :

df1 :

date_1      value_1
2017-1-15    20
2017-1-31    30
2016-2-15    20

df2:

date_2      value_2
2017-1-1    30

then get result as below i.e. merged on the basis of month and year.

date_1      value_1 value_2
2017-1-15    20     30
2017-1-31    30     30
2016-2-15    20

Here's a rudimentary but effective solution:

res = pd.merge(df1.assign(grouper=df1['date_1'].dt.to_period('M')),df2.assign(grouper=df2['date_2'].dt.to_period('M')),how='left', on='grouper')print(res)date_1  value_1 grouper     date_2  value_2
0 2017-01-15       20 2017-01 2017-01-01     30.0
1 2017-01-31       30 2017-01 2017-01-01     30.0
2 2016-02-15       20 2016-02        NaT      NaN

You can then remove unwanted columns:

res = res[['date_1', 'value_1', 'value_2']]