0

I have a CSV with some information about products color. Since sometimes some extra details are there, I would like to extract just the color name. I found out some library but my data are in french so they dont fit those. I try to do it with Python.

From "transparent blue" I want to just keep "blue"

The table is like :

Product ref Color Sales quantity
F33 Bleu transparent 2
K367 Ecaille Marron 1

And I am looking to take the "Bleu" (Blue) and "Marron" (brown) to see which colors are the more sale

Vivendi
  • 20,047
  • 25
  • 121
  • 196
Marie A
  • 3
  • 2
  • What programming language are you using? Can you also provide a more detailed example of the CSV? – Vivendi Jan 25 '23 at 08:53

1 Answers1

0

You could create a translator function and then apply this to the column.

here is an example (using the data in the question).

import pandas as pd

# original dataframe
data = {'Product ref': ['F33', 'K367'],
        'Color': ['Bleu transparent', 'Ecaille Marron'],
        'Sales quantity': [2, 1]}

df = pd.DataFrame(data)


def translate(french):
    ''' translating function '''
    if 'Bleu' in french:
        return 'blue'
    
    if 'Marron' in french:
        return 'brown'
    
    return '-'

# apply the result
df['english'] = df['Color'].apply(translate)
print(df)

This is the result:

  Product ref             Color  Sales quantity english
0         F33  Bleu transparent               2    blue
1        K367    Ecaille Marron               1  brown

Note: You could use a much more sophistocated translating and matching function (for example googletrans). The example above is a working example.

imM4TT
  • 292
  • 1
  • 8
D.L
  • 4,339
  • 5
  • 22
  • 45
  • What i am scare about with this idea is there are at least 40 different colors ... then the translate fonction would be pretty long ... – Marie A Jan 26 '23 at 14:24
  • @MarieA, yes, it would, which is why i showed only how to implement the "translator" function correctly (and efficiently). The reality is that you would use pattern matching & translation, but this would be the subject of another question. – D.L Jan 26 '23 at 15:54
  • @MarieA, a quick (and also computational fast) solution to having many colours would be to implement a hash table (python dictionary) in place of the if statements that i used in the example... it is relatively simple to implement. – D.L Jan 26 '23 at 15:58