How to assign a value to a column for every row of pandas dataframe?

Question

I've been trying to assign a value for every row of a dataframe and I haven't been able to do so (I'm new in pandas), so if anyone could help, I'd be super grateful!

I've got two dataframes. In the input dataframe, I have brands:

brand_raw.head()

brand_name
0   Nike
1   Lacoste
2   Adidas

And then, on the output dataset, I have objects:

object_raw.head()

category_id object_name
0   24  T-shirt
1   45  Shorts
2   32  Dress

and what I would need to have is a dataframe with all the objects combined with all the brands:

to_raw.head()

category_id object_name brand_name
0   24  T-shirt     Nike
1   45  Shorts      Nike
2   32  Dress       Nike
3   24  T-shirt     Lacoste
4   45  Shorts      Lacoste
5   32  Dress       Lacoste
6   24  T-shirt     Adidas
7   45  Shorts      Adidas
8   32  Dress       Adidas

I've been trying to do it with the apply function, iterating over the rows, but I end up overwriting the values so I write the last brand:

0   24  T-shirt     Nike
1   45  Shorts      Nike
2   32  Dress       Nike

This is my code:

def insert_value_in_every_row(input_df, output_df, column_name):
    for row in input_df.values:
        row = row[0].rstrip()

        output_df[column_name] = output_df[column_name].apply(lambda x: row) 
    return output_df


insert_value_in_every_row(brand_raw, to_raw, 'brand_name')

Could someone give me a hint on how to deal with this, please? Thanks a lot in advance!

You can try and take a look at this ? https://stackoverflow.com/questions/13269890/cartesian-product-in-pandas — Born Tbe Wasted, Mar 29 '19 at 11:39

yatu · Accepted Answer · 2019-03-29T11:48:35.690

You're looking for a cartesian product of both dataframes. One way around this in pandas is to create a common and unique key for both dataframes and perform a merge (any, as there is a complete overlap):

df.assign(key=0).merge(object_raw.assign(key=0), on='key').drop(['key'], axis=1)

   brand_name  category_id object_name
0       Nike           24     T-shirt
1       Nike           45      Shorts
2       Nike           32       Dress
3    Lacoste           24     T-shirt
4    Lacoste           45      Shorts
5    Lacoste           32       Dress
6     Adidas           24     T-shirt
7     Adidas           45      Shorts
8     Adidas           32       Dress

score 2 · Answer 2 · answered Mar 29 '19 at 11:49

Another way using itertools.product() which gives Cartesian product of input iterables.

import itertools
df=(pd.DataFrame(list(itertools.product(brand_name.brand_name,object_raw.object_name))
             ,columns=['brand_name','object_name']))
df['category_id']=df['object_name'].map(object_raw.set_index('object_name')['category_id'])
print(df)

  brand_name object_name  category_id
0       Nike     T-shirt           24
1       Nike      Shorts           45
2       Nike       Dress           32
3    Lacoste     T-shirt           24
4    Lacoste      Shorts           45
5    Lacoste       Dress           32
6     Adidas     T-shirt           24
7     Adidas      Shorts           45
8     Adidas       Dress           32

score 1 · Answer 3 · answered Mar 29 '19 at 11:41

This is known as a cartesian product. In pandas its a bit tricky, but you could do it like this:

import pandas as pd

objects = pd.DataFrame(["T-shirt","Shorts","Dress"],columns = ['object'])
brands =  pd.DataFrame(["Nike","Lacoste","Adidas"],columns = ['brand'])

objects["key"] = 1
brands ["key"] = 1

objects.merge(brands,on='key').drop('key',axis=1)

How to assign a value to a column for every row of pandas dataframe?

3 Answers3