使用 for 循环替换 pandas 列的每一行中的单元格值

Question

Please, help me understand my error. I'm trying to change one column in my .csv file. I have .csv file as following:

sku,name,code  
k1,aaa,886  
k2,bbb,898  
k3,ccc,342  
k4,ddd,503  
k5,eee,401  

I want to replace "k" symbol with the "_" symbol in the "sku" column.
I wrote the code:

import sys  
import pandas as pd  
import numpy as np  
import datetime  

df = pd.read_csv('cat0.csv')  

for r in df['sku']:  
    r1 = r.replace('k', '_')  
    df['sku'] = r1  

print (df) 

But the code inserts the last value in every row of the "sku" column. So I get:

  sku name  code
0  _5  aaa   886
1  _5  bbb   898
2  _5  ccc   342
3  _5  ddd   503
4  _5  eee   401

I want to get as following:

  sku name  code
0  _1  aaa   886
1  _2  bbb   898
2  _3  ccc   342
3  _4  ddd   503
4  _5  eee   401

Answer 1

You can use str.replace on the whole column:

from io import StringIO
import pandas as pd

data = """sku,name,code  
k1,aaa,886  
k2,bbb,898  
k3,ccc,342  
k4,ddd,503  
k5,eee,401"""

file = StringIO(data)

df = pd.read_csv(file)
df['sku'] = df['sku'].str.replace('k', '_')

print(df)

This yields

  sku name  code  
0  _1  aaa     886
1  _2  bbb     898
2  _3  ccc     342
3  _4  ddd     503
4  _5  eee     401

Answer 2

As @Jan mentioned, doing it by using df['sku'] = df['sku'].str.replace('k', '_') is the best/quickest way to do this.

However, to understand why you are getting the results you are and to present a way as close to how you were doing it as possible, you'd do:

import pandas as pd

df = pd.DataFrame(
    {
        'sku':["k1", "k2", "k3", "k4", "k5"], 
        'name': ["aaa", "bbb", "ccc", "ddd", "eee"], 
        'code':[886, 898,342,503,401]
    }, columns =["sku", "name", "code"]
)

for i, r in enumerate(df['sku']):  
    r1 = r.replace('k', '_')
    df.at[i, 'sku'] = r1  

Which gives:

  sku name  code
0  _1  aaa   886
1  _2  bbb   898
2  _3  ccc   342
3  _4  ddd   503
4  _5  eee   401

In your code...

for r in df['sku']:  
    r1 = r.replace('k', '_')  

...the issue is here:

    df['sku'] = r1  

You are broadcasting your results to the entire column rather than just the row you are working on.