Creating Dummy Variables from String Column

Question

I have a pandas dataframe (N = 1485) that looks like this:

ID          Intervention
1           Blood Draw, Flushed, Locked
1           Blood Draw, Port De-Accessed, Heparin-Locked, Tubing Changed
1           Blood Draw, Flushed
2           Blood return Verified, Flushed
2           Cap Changed
3           Port De-Accessed

I want to be able to dummy code out each of the string before every comma so it looks similar to this:

ID          Blood Draw          Flushed          Locked      ....
1              Yes                Yes             Yes
1              Yes                No              No
...

Thanks!

look at `str.get_dummies` https://pandas.pydata.org/pandas-docs/stable/generated/pandas.Series.str.get_dummies.html — BENY, Aug 17 '18 at 18:35
Also search on "one-hot encoding". Pandas has a one-hot operation. — Prune, Aug 17 '18 at 18:37

score 3 · Accepted Answer · answered Aug 17 '18 at 19:55

You can use pd.Series.str.get_dummies and a dictionary mapping:

d = {1: 'yes', 0: 'no'}
res = df.join(df.pop('Intervention').str.get_dummies(', ').applymap(d.get))

In my opinion, it's best to convert to strings for display purposes only. Boolean values are more efficiently held and manipulated in Boolean series.

Result

print(res)

   ID Blood Draw Blood return Verified Cap Changed Flushed Heparin-Locked  \
0   1        yes                    no          no     yes             no   
1   1        yes                    no          no      no            yes   
2   1        yes                    no          no     yes             no   
3   2         no                   yes          no     yes             no   
4   2         no                    no         yes      no             no   
5   3         no                    no          no      no             no   

  Locked Port De-Accessed Tubing Changed  
0    yes               no             no  
1     no              yes            yes  
2     no               no             no  
3     no               no             no  
4     no               no             no  
5     no              yes             no

Setup

df = pd.DataFrame({'ID': [1, 1, 1, 2, 2, 3],
                   'Intervention': ['Blood Draw, Flushed, Locked',
                                    'Blood Draw, Port De-Accessed, Heparin-Locked, Tubing Changed',
                                    'Blood Draw, Flushed', 'Blood return Verified, Flushed',
                                    'Cap Changed', 'Port De-Accessed']})

score 0 · Answer 2 · answered Aug 17 '18 at 18:38

0

You could try the following:

for event in ['Blood Draw', 'Flushed', 'Locked']:
    df[event] = df['Intervention'].str.contains(event)

which will give you True/False instead of 'Yes'/'No' which may be more useful when you are post-processing.

answered Aug 17 '18 at 18:38

jeschwar

1,286
7
10

score 0 · Answer 3 · answered Aug 17 '18 at 19:33

import numpy as np
df1=df['Intervention'].str.split(',', expand=True)  
df2=df1.replace(np.nan, '', regex=True) # Replacing None with blank data
pd.concat([pd.get_dummies(df2[col]) for col in df2], axis=1, keys=df2.columns)  # Creates dummies for all the columns

To perform the above steps, filter the Intervention column, do this process and join with original dataframe so that the dummies statement works ( Creates the dummies for all the columns).

Let me know if this is what you wanted. – Harikrishna Aug 17 '18 at 19:35 — Harikrishna, Aug 17 '18 at 19:35

Creating Dummy Variables from String Column

3 Answers3