Pandas 比较同一数据框中两列中的字符串，并有条件地输出到新列

Question

I have two columns within a data frame containing strings. For example,

import pandas as pd
import numpy as np

data = [['Oct-2019', 'Oranges + Grapes + Pears', 'Grapes + Pears'],
       ['Nov-2019', 'Oranges + Grapes + Pears', 'Oranges + Grapes + Pears']]

df = pd.DataFrame(data, columns =['Date', 'Previous shopping list', 'Recent shopping list'])
print(df)

Fish = ['Salmon', 'Trout']
Fruit = ['Oranges', 'Grapes', 'Pears']

     Date     PSL                 RSL
0  Oct-2019   Oranges + Grapes    Grapes + Pears
              + Pears + Salmon                     

1  Nov-2019   Oranges + Grapes    Oranges + Grapes
              + Pears + Trout     + Pears  

I want to compare the strings in both columns and have a text output to a new column that says what has changed between the two lists. Such as, creating a column that will check for the strings related to "Fruit" and output what fruit has been dropped from the recent shopping when compared to the previous list previous shopping list. See Desired output below:

     Date     PSL                 RSL               Fruit lost   Fish Lost
0  Oct-2019   Oranges + Grapes    Grapes + Pears    Oranges      Salmon
              + Pears + Salmon                     

1  Nov-2019   Oranges + Grapes    Oranges + Grapes               Trout
              + Pears + Trout     + Pears  

How would I be able to achieve this in using pandas! Apologies if this was not clear the first time!

Thank you for any suggestion/help!

Answer 1

The exact function that you use to process the data depends on your exact output that you require for each combination. Hopefully below will give you enough to create a solution for your problem:

# process data so each row contains a list of elements
df['PSL_processed'] = df['Previous shopping list'].str.split('+')
df['RSL_processed'] = df['Recent shopping list'].str.split('+')

def compare_items(x):
    if set(x.PSL_processed) == set(x.RSL_processed):
        return 'No change'
    elif set(x.PSL_processed) - set(x.CSL_processed) > 0:
        return 'Lost'
    # add in conditional logic here, to meet specification

df.apply(compare_items, axis=1)

The official documentation for pd.apply() is well written.

Answer 2

要检查“最近的购物清单”中是否存在字符串“Oranges”并根据结果创建一个新列“Oranges Lost”：

df['Oranges Lost'] = np.where(df['Recent shopping list'].str.contains('Oranges'), 'No Change', 'Lost')``` 

Answer 3

So Mark's solution works well to grab the difference between the lists

# process data so each row contains a list of elements
df['PSL_processed'] = df['Previous shopping list'].str.split()
df['RSL_processed'] = df['Recent shopping list'].str.split()

def compare_items(x):
    return set(x.PSL_processed) - set(x.RSL_processed)
    # add in conditional logic here, to meet specification
df['Products_lost'] = df.apply(compare_items, axis=1)

print(df)

On top to that to find the products that = fruit and the products = fish I used the following:

for idx, row in df.iterrows():
    for c in Fruit:
        if c in row['Products_lost']:
            df.ix[idx, 'Fruit lost'] = c
            for c in Fish:
                if c in row['Products_lost']:
                    df.ix[idx, 'Fish lost'] = c

Seems to work well!