通过for循环更新pandas数据帧

Question

I have a bunch of URLs stored into a data frame and I am cleaning them up via a url parsing module. The issue that I am having is that the 'siteClean' field that is supposed to update with the cleaned url is updating the entire column and not the individual cell...

Here is the code:

results = resultsX.copy(deep = True)
results = results.reset_index(drop = True)
results['siteClean'] = ''


from urlparse import urlsplit  
import re

for row in results.iterrows():
    #print row[1]
    url = row[1][1]
    if not re.match(r'http(s?)\:', url):
        url = 'http://' + url
    parsed = urlsplit(url)
    host = parsed.netloc
    #print host
    #row[1][1] = host
    #results[row][1] = host
    results['siteClean'] = host
    print results

Answer 1

In general, it's better to avoid looping over your frame's rows, if you can avoid it. If I understand your problem correctly, you want to look at a single column from your frame, and apply a function on each element of that column. Then you want to put the result of all those function calls into a column of the original frame. Maybe a new column, maybe in place of the old column. This sounds like a job for pd.Series.map .

import pandas as pd
import numpy as np

np.random.seed(0)

n=10

df = pd.DataFrame({'num': np.random.randn(n),
                   'lett': np.random.choice(
                        list('abcdefghijklmnopqrstuvwxyz'),n)
                   })  

df looks like this:

Set up a function to classify a single letter as either a consonant or a vowel:

def classify_letter(char):
    if char in list('aeiou'):
        return 'vowel'
    else:
        return 'consonant'

Then you can use map to generate a new Series whose entries are those of the input transformed by the specified function. You can stick that new output series wherever you like. It can be a new column (in your old DataFrame or elsewhere) or it can replace the old column. Note that map only works on a Series , so be sure to select down to one column before using it:

df['new'] = df['lett'].map(classify_letter)

gives:

while if you started from the original setup and ran:

df['lett'] = df['lett'].map(classify_letter)

then you would replace the old column with the new one: