pandas dataframe extract strings

Question

My data frame has a column called 'a' and it may contain 'apple' and 'orange'. What I want is to extract them if they exist, otherwise label 'others'.

I can simply loop over the rows and extract them. However, I saw some usage of numpy.where() for similar purposes, but only two categories.

result = numpy.where(df['a'].str.contains('apple'), 'apple', 'others')

Is it possible to apply it here for the case of 3 categories? In other words, result should contain entries of 'apple', 'orange', or 'others'.

Is there some better way to do it than simply looping?

Community · Accepted Answer · 2017-05-23 12:22:30Z

Simply look for items that are apple or mango with np.in1d to create a boolean mask, which could then be used with np.where to set rest of them as others. Thus, we would have -

df['b'] = np.where(np.in1d(df.a,['apple','orange']),df.a,'others')

For cases when you might be looking to work with strings that have those names as part of bigger strings, you can use str.extract (caught this idea from @jezrael's solution, I hope that's okay!) and then use np.where, like so -

strings = df.a.str.extract('(orange|apple)')
df['b'] = np.where(np.in1d(strings,['apple','orange']),strings,'others')

Sample run -

In [294]: df
Out[294]: 
             a
0  apple-shake
1       orange
2  apple-juice
3        apple
4        mango
5       orange
6       banana

In [295]: strings = df.a.str.extract('(orange|apple)')

In [296]: df['b'] = np.where(np.in1d(strings,['apple','orange']),strings,'others')

In [297]: df
Out[297]: 
             a       b
0  apple-shake   apple
1       orange  orange
2  apple-juice   apple
3        apple   apple
4        mango  others
5       orange  orange
6       banana  others

jezrael · Accepted Answer · 2016-07-12 21:05:23Z

2

Use str.extract with fillna:

df = pd.DataFrame({'a': ['orange','apple','a']})
print (df)
        a
0  orange
1   apple
2       a

df['new'] = df.a.str.extract('(orange|apple)', expand=False).fillna('others')
print (df)
        a     new
0  orange  orange
1   apple   apple
2       a  others

edited Jul 12, 2016 at 21:05

answered Jul 12, 2016 at 21:02

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

1 Comment

nos Over a year ago

I want the result to be one of the 3 possibilities: 'apple', 'orange' or 'others'.

Collectives™ on Stack Overflow

pandas dataframe extract strings

2 Answers 2

Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related