Pandas add new column based on another column

Question

I have a List of list values as shown below:

res = [["a", "b", "b"], ["d", "e"], ["f", "g"]]

I have a data frame as shown below:

df__ = pd.DataFrame({'memberid': ['a1','a2','a3'],
                   'labels':[0,1,2]})

It'll look as shown below:

              labels memberid
        0       0       a1
        1       1       a2
        2       2       a3

I want to add another column called prob based on the labelscolumn, the value constituting from res list. The output will look as what you get when you run the below:

df__ = pd.DataFrame({'memberid': ['a1','a2','a3'],
                   'labels':[0,1,2],


       labels memberid   prob
0       0       a1       a b b
1       1       a2       d e
2       2       a3       f g

So basically, I use the labels value as the index for res list and populate the prob column.

I have run the code below:

for i in range(len(df__["labels"])):
           k =  df__.iloc[i]["labels"]
           df__["prob"] = " ".join(res[k])

But I don't get the output that I want from the above code. What am I doing wrong?

please see my answer. The problem is, you keep reassigning df__["prob"] under the for/loop hence the last value wins — Anzel
– Anzel, Commented Nov 10, 2014 at 11:48

Anzel · Accepted Answer · 2014-11-10 11:38:48Z

3

Re your error, it lies on this:

df__["prob"] = " ".join(res[k])

You keep reassigning df__["prob"] = 1 value, which is the latest " ".join(res[l]) Hence at the end the whole column is just the last value. To correct this, you can change to this:

prob = []
for i in range(len(df__["labels"])):
    k =  df__.iloc[i]["labels"]
    prob.append(" ".join(res[k]))
df__['prob'] = prob

Also you can use map and lambda, like this to achieve the same result, this is more efficient than your attempt:

import pandas as pd

df__ = pd.DataFrame({'memberid': ['a1','a2','a3'],
                   'labels':[0,1,2]})
res = [["a", "b", "b"], ["d", "e"], ["f", "g"]]
# you can map the values from '__labels' and feed to 'prob' with lambda
df__['prob'] = map(lambda x: ' '.join(res[x]), df__['labels'])

df__
   labels memberid   prob
0       0       a1  a b b
1       1       a2    d e
2       2       a3    f g

edited Nov 10, 2014 at 11:38

answered Nov 10, 2014 at 11:24

Anzel

20.6k5 gold badges54 silver badges53 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

user1452759 Over a year ago

Thank you so much for this!!! :) It works! And thank you for showing the pandas way to do it too.

Anzel Over a year ago

@user1452759, not a problem! glad it helps :)

Collectives™ on Stack Overflow

Pandas add new column based on another column

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related