changing values in a column of data set using regex in pandas

Question

This is a subset of a data frame:

Index     duration 
1          4  months20mg 1X D
2          1  years10 1X D
3          2  weeks10 mg
4          8  years300 MG 1X D
5          20  days
6          10  months

The output should be like this:

Index     duration 
1          4  month
2          1  year
3          2  week
4          8  year
5          20  day
6          10  month

This is my code:

df.dosage_duration.replace(r'year[0-9a-zA-z]*' , 'year', regex=True)
df.dosage_duration.replace(r'day[0-9a-zA-z]*' , 'day', regex=True)
df.dosage_duration.replace(r'month[0-9a-zA-z]*' , 'month', regex=True)
df.dosage_duration.replace(r'week[0-9a-zA-z]*' , 'week', regex=True)

But it does not work. Any suggestion ?

df.duration.str.replace('((?<=year)|(?<=month)|(?<=week)|(?<=day)).*', '') — Ted Petrou
– Ted Petrou, Commented Jun 28, 2017 at 4:09

chthonicdaemon · Accepted Answer · 2017-06-28 04:07:32Z

3

There are two problems.

The first is that your regular expression doesn't match all the parts you want it to match. Look at months20mg 1X D - there is a space in the part you want to replace. I think you could probably just use 'year.*' as your matches.

The second is that you are calling replace without storing the results. If you want to do the call the way you have, you should specify inplace=True.

You can also use a single call if you use a slightly extended regular expression. We can use \1 to refer to the first matching group for the regular expression. The groups are indicated by the parentheses:

df.dosage_duration.replace(r'(year|month|week|day).*' , r'\1', 
                           regex=True, inplace=True)

edited Jun 28, 2017 at 4:07

answered Jun 28, 2017 at 4:00

chthonicdaemon

19.9k2 gold badges55 silver badges70 bronze badges

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

5 Comments

Mary Over a year ago

thank you. Would you please explain how r'\1' works ?

0xffcourse Over a year ago

Also, if it is guarrented that the day,week,month,week end with an s, I guess, you can simply do this df.dosage_duration.split('s')[0].

chthonicdaemon Over a year ago

@Mary I've added to my answer.

Mary Over a year ago

@officialaimm, what is the meaning of [0] at the end of your code ?

0xffcourse Over a year ago

Take first(0-indexed) half from the split i.e. the part left to the 's'. split() returns an array ['4 month', '20mg 1X D']. Now the 0th element of the array is what we are interested in.

Collectives™ on Stack Overflow

changing values in a column of data set using regex in pandas

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related