Julia Bogutskaya Julia Bogutskaya - 1 month ago 11
Python Question

Adding Dates (Series) column from one DataFrame to the other Pandas, Python

I am trying to 'broadcast' a date column from df1 to df2.

In df1 I have the names of all the users and their basic information.
In df2 I have a list of purchases made by the users.

df1 and df2 code

Assuming I have a much bigger dataset (the above created for sample) how can I add just(!) the df1['DoB'] column to df2?

I have tried both concat() and merge() but none of them seem to work:

code and error

The only way it seems to work is only if I merge both df1 and df2 together and then just delete the columns I don't need. But if I have tens of unwanted columns, it is going to be very problematic.

The full code (including the lines that throw an error):

import pandas as pd
df1 = pd.DataFrame(columns=['Name','Age','DoB','HomeTown'])

df1['Name'] = ['John', 'Jack', 'Wendy','Paul']
df1['Age'] = [25,23,30,31]
df1['DoB'] = pd.to_datetime(['04-01-2012', '03-02-1991', '04-10-1986', '06-03-1985'], dayfirst=True)
df1['HomeTown'] = ['London', 'Brighton', 'Manchester', 'Jersey']

df2 = pd.DataFrame(columns=['Name','Purchase'])
df2['Name'] = ['John','Wendy','John','Jack','Wendy','Jack','John','John']
df2['Purchase'] = ['fridge','coffee','washingmachine','tickets','iPhone','stove','notebook','laptop']

df2 = df2.concat(df1) # error

df2 = df2.merge(df1['DoB'], on='Name', how='left') #error

df2 = df2.merge(df1, on='Name', how='left')
del df2['Age'], df2['HomeTown']
df2 #that's how i want it to look like


Any help would be much appreciated. Thank you :)



Answer

I think you need merge with subset [['Name','DoB']] - need Name column for matching:

print (df1[['Name','DoB']])
    Name        DoB
0   John 2012-01-04
1   Jack 1991-02-03
2  Wendy 1986-10-04
3   Paul 1985-03-06

df2 = df2.merge(df1[['Name','DoB']], on='Name', how='left')
print (df2)
    Name        Purchase        DoB
0   John          fridge 2012-01-04
1  Wendy          coffee 1986-10-04
2   John  washingmachine 2012-01-04
3   Jack         tickets 1991-02-03
4  Wendy          iPhone 1986-10-04
5   Jack           stove 1991-02-03
6   John        notebook 2012-01-04
7   John          laptop 2012-01-04

Another solution with map by Series s:

s = df1.set_index('Name')['DoB']
print (s)
Name
John    2012-01-04
Jack    1991-02-03
Wendy   1986-10-04
Paul    1985-03-06
Name: DoB, dtype: datetime64[ns]

df2['DoB'] = df2.Name.map(s)
print (df2)
    Name        Purchase        DoB
0   John          fridge 2012-01-04
1  Wendy          coffee 1986-10-04
2   John  washingmachine 2012-01-04
3   Jack         tickets 1991-02-03
4  Wendy          iPhone 1986-10-04
5   Jack           stove 1991-02-03
6   John        notebook 2012-01-04
7   John          laptop 2012-01-04
Comments