Working with Text Data

Series and Index are equipped with a set of string processing methods that make it easy to operate on each element of the array. Perhaps most importantly, these methods exclude missing/NA values automatically. These are accessed via the str attribute and generally have names matching the equivalent (scalar) built-in string methods:

In [1]: s = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca', np.nan, 'CABA', 'dog', 'cat'])

In [2]: s.str.lower()
Out[2]: 
0       a
1       b
2       c
3    aaba
4    baca
5     NaN
6    caba
7     dog
8     cat
dtype: object

In [3]: s.str.upper()
Out[3]: 
0       A
1       B
2       C
3    AABA
4    BACA
5     NaN
6    CABA
7     DOG
8     CAT
dtype: object

In [4]: s.str.len()
Out[4]: 
0    1.0
1    1.0
2    1.0
3    4.0
4    4.0
5    NaN
6    4.0
7    3.0
8    3.0
dtype: float64
In [5]: idx = pd.Index([' jack', 'jill ', ' jesse ', 'frank'])

In [6]: idx.str.strip()
Out[6]: Index([u'jack', u'jill', u'jesse', u'frank'], dtype='object')

In [7]: idx.str.lstrip()
Out[7]: Index([u'jack', u'jill ', u'jesse ', u'frank'], dtype='object')

In [8]: idx.str.rstrip()
Out[8]: Index([u' jack', u'jill', u' jesse', u'frank'], dtype='object')

The string methods on Index are especially useful for cleaning up or transforming DataFrame columns. For instance, you may have columns with leading or trailing whitespace:

In [9]: df = pd.DataFrame(randn(3, 2), columns=[' Column A ', ' Column B '],
   ...:                   index=range(3))
   ...: 

In [10]: df
Out[10]: 
    Column A    Column B 
0    0.017428    0.039049
1   -2.240248    0.847859
2   -1.342107    0.368828

Since df.columns is an Index object, we can use the .str accessor

In [11]: df.columns.str.strip()
Out[11]: Index([u'Column A', u'Column B'], dtype='object')

In [12]: df.columns.str.lower()
Out[12]: Index([u' column a ', u' column b '], dtype='object')

These string methods can then be used to clean up the columns as needed. Here we are removing leading and trailing whitespaces, lowercasing all names, and replacing any remaining whitespaces with underscores:

In [13]: df.columns = df.columns.str.strip().str.lower().str.replace(' ', '_')

In [14]: df
Out[14]: 
   column_a  column_b
0  0.017428  0.039049
1 -2.240248  0.847859
2 -1.342107  0.368828

Note

If you have a Series where lots of elements are repeated (i.e. the number of unique elements in the Series is a lot smaller than the length of the Series), it can be faster to convert the original Series to one of type category and then use .str.<method> or .dt.<property> on that. The performance difference comes from the fact that, for Series of type category, the string operations are done on the .categories and not on each element of the Series.

Please note that a Series of type category with string .categories has some limitations in comparison of Series of type string (e.g. you can’t add strings to each other: s + " " + s won’t work if s is a Series of type category). Also, .str methods which operate on elements of type list are not available on such a Series.

Splitting and Replacing Strings

Methods like split return a Series of lists:

In [15]: s2 = pd.Series(['a_b_c', 'c_d_e', np.nan, 'f_g_h'])

In [16]: s2.str.split('_')
Out[16]: 
0    [a, b, c]
1    [c, d, e]
2          NaN
3    [f, g, h]
dtype: object

Elements in the split lists can be accessed using get or [] notation:

In [17]: s2.str.split('_').str.get(1)
Out[17]: 
0      b
1      d
2    NaN
3      g
dtype: object

In [18]: s2.str.split('_').str[1]
Out[18]: 
0      b
1      d
2    NaN
3      g
dtype: object

Easy to expand this to return a DataFrame using expand.

In [19]: s2.str.split('_', expand=True)
Out[19]: 
     0     1     2
0    a     b     c
1    c     d     e
2  NaN  None  None
3    f     g     h

It is also possible to limit the number of splits:

In [20]: s2.str.split('_', expand=True, n=1)
Out[20]: 
     0     1
0    a   b_c
1    c   d_e
2  NaN  None
3    f   g_h

rsplit is similar to split except it works in the reverse direction, i.e., from the end of the string to the beginning of the string:

In [21]: s2.str.rsplit('_', expand=True, n=1)
Out[21]: 
     0     1
0  a_b     c
1  c_d     e
2  NaN  None
3  f_g     h

Methods like replace and findall take regular expressions, too:

In [22]: s3 = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca',
   ....:                '', np.nan, 'CABA', 'dog', 'cat'])
   ....: 

In [23]: s3
Out[23]: 
0       A
1       B
2       C
3    Aaba
4    Baca
5        
6     NaN
7    CABA
8     dog
9     cat
dtype: object

In [24]: s3.str.replace('^.a|dog', 'XX-XX ', case=False)
Out[24]: 
0           A
1           B
2           C
3    XX-XX ba
4    XX-XX ca
5            
6         NaN
7    XX-XX BA
8      XX-XX 
9     XX-XX t
dtype: object

Some caution must be taken to keep regular expressions in mind! For example, the following code will cause trouble because of the regular expression meaning of $:

# Consider the following badly formatted financial data
In [25]: dollars = pd.Series(['12', '-$10', '$10,000'])

# This does what you'd naively expect:
In [26]: dollars.str.replace('$', '')
Out[26]: 
0        12
1       -10
2    10,000
dtype: object

# But this doesn't:
In [27]: dollars.str.replace('-$', '-')
Out[27]: 
0         12
1       -$10
2    $10,000
dtype: object

# We need to escape the special character (for >1 len patterns)
In [28]: dollars.str.replace(r'-\$', '-')
Out[28]: 
0         12
1        -10
2    $10,000
dtype: object

Indexing with .str

You can use [] notation to directly index by position locations. If you index past the end of the string, the result will be a NaN.

In [29]: s = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca', np.nan,
   ....:                'CABA', 'dog', 'cat'])
   ....: 

In [30]: s.str[0]
Out[30]: 
0      A
1      B
2      C
3      A
4      B
5    NaN
6      C
7      d
8      c
dtype: object

In [31]: s.str[1]
Out[31]: 
0    NaN
1    NaN
2    NaN
3      a
4      a
5    NaN
6      A
7      o
8      a
dtype: object

Extracting Substrings

Extract first match in each subject (extract)

New in version 0.13.0.

Warning

In version 0.18.0, extract gained the expand argument. When expand=False it returns a Series, Index, or DataFrame, depending on the subject and regular expression pattern (same behavior as pre-0.18.0). When expand=True it always returns a DataFrame, which is more consistent and less confusing from the perspective of a user.

The extract method accepts a regular expression with at least one capture group.

Extracting a regular expression with more than one group returns a DataFrame with one column per group.

In [32]: pd.Series(['a1', 'b2', 'c3']).str.extract('([ab])(\d)', expand=False)
Out[32]: 
     0    1
0    a    1
1    b    2
2  NaN  NaN

Elements that do not match return a row filled with NaN. Thus, a Series of messy strings can be “converted” into a like-indexed Series or DataFrame of cleaned-up or more useful strings, without necessitating get() to access tuples or re.match objects. The dtype of the result is always object, even if no match is found and the result only contains NaN.

Named groups like

In [33]: pd.Series(['a1', 'b2', 'c3']).str.extract('(?P<letter>[ab])(?P<digit>\d)', expand=False)
Out[33]: 
  letter digit
0      a     1
1      b     2
2    NaN   NaN

and optional groups like

In [34]: pd.Series(['a1', 'b2', '3']).str.extract('([ab])?(\d)', expand=False)
Out[34]: 
     0  1
0    a  1
1    b  2
2  NaN  3

can also be used. Note that any capture group names in the regular expression will be used for column names; otherwise capture group numbers will be used.

Extracting a regular expression with one group returns a DataFrame with one column if expand=True.

In [35]: pd.Series(['a1', 'b2', 'c3']).str.extract('[ab](\d)', expand=True)
Out[35]: 
     0
0    1
1    2
2  NaN

It returns a Series if expand=False.

In [36]: pd.Series(['a1', 'b2', 'c3']).str.extract('[ab](\d)', expand=False)
Out[36]: 
0      1
1      2
2    NaN
dtype: object

Calling on an Index with a regex with exactly one capture group returns a DataFrame with one column if expand=True,

In [37]: s = pd.Series(["a1", "b2", "c3"], ["A11", "B22", "C33"])

In [38]: s
Out[38]: 
A11    a1
B22    b2
C33    c3
dtype: object

In [39]: s.index.str.extract("(?P<letter>[a-zA-Z])", expand=True)
Out[39]: 
  letter
0      A
1      B
2      C

It returns an Index if expand=False.

In [40]: s.index.str.extract("(?P<letter>[a-zA-Z])", expand=False)
Out[40]: Index([u'A', u'B', u'C'], dtype='object', name=u'letter')

Calling on an Index with a regex with more than one capture group returns a DataFrame if expand=True.

In [41]: s.index.str.extract("(?P<letter>[a-zA-Z])([0-9]+)", expand=True)
Out[41]: 
  letter   1
0      A  11
1      B  22
2      C  33

It raises ValueError if expand=False.

>>> s.index.str.extract("(?P<letter>[a-zA-Z])([0-9]+)", expand=False)
ValueError: only one regex group is supported with Index

The table below summarizes the behavior of extract(expand=False) (input subject in first column, number of groups in regex in first row)

  1 group >1 group
Index Index ValueError
Series Series DataFrame

Extract all matches in each subject (extractall)

New in version 0.18.0.

Unlike extract (which returns only the first match),

In [42]: s = pd.Series(["a1a2", "b1", "c1"], ["A", "B", "C"])

In [43]: s
Out[43]: 
A    a1a2
B      b1
C      c1
dtype: object

In [44]: two_groups = '(?P<letter>[a-z])(?P<digit>[0-9])'

In [45]: s.str.extract(two_groups, expand=True)
Out[45]: 
  letter digit
A      a     1
B      b     1
C      c     1

the extractall method returns every match. The result of extractall is always a DataFrame with a MultiIndex on its rows. The last level of the MultiIndex is named match and indicates the order in the subject.

In [46]: s.str.extractall(two_groups)
Out[46]: 
        letter digit
  match             
A 0          a     1
  1          a     2
B 0          b     1
C 0          c     1

When each subject string in the Series has exactly one match,

In [47]: s = pd.Series(['a3', 'b3', 'c2'])

In [48]: s
Out[48]: 
0    a3
1    b3
2    c2
dtype: object

then extractall(pat).xs(0, level='match') gives the same result as extract(pat).

In [49]: extract_result = s.str.extract(two_groups, expand=True)

In [50]: extract_result
Out[50]: 
  letter digit
0      a     3
1      b     3
2      c     2

In [51]: extractall_result = s.str.extractall(two_groups)

In [52]: extractall_result
Out[52]: 
        letter digit
  match             
0 0          a     3
1 0          b     3
2 0          c     2

In [53]: extractall_result.xs(0, level="match")
Out[53]: 
  letter digit
0      a     3
1      b     3
2      c     2

Testing for Strings that Match or Contain a Pattern

You can check whether elements contain a pattern:

In [54]: pattern = r'[a-z][0-9]'

In [55]: pd.Series(['1', '2', '3a', '3b', '03c']).str.contains(pattern)
Out[55]: 
0    False
1    False
2    False
3    False
4    False
dtype: bool

or match a pattern:

In [56]: pd.Series(['1', '2', '3a', '3b', '03c']).str.match(pattern, as_indexer=True)
Out[56]: 
0    False
1    False
2    False
3    False
4    False
dtype: bool

The distinction between match and contains is strictness: match relies on strict re.match, while contains relies on re.search.

Warning

In previous versions, match was for extracting groups, returning a not-so-convenient Series of tuples. The new method extract (described in the previous section) is now preferred.

This old, deprecated behavior of match is still the default. As demonstrated above, use the new behavior by setting as_indexer=True. In this mode, match is analogous to contains, returning a boolean Series. The new behavior will become the default behavior in a future release.

Methods like match, contains, startswith, and endswith take
an extra na argument so missing values can be considered True or False:
In [57]: s4 = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca', np.nan, 'CABA', 'dog', 'cat'])

In [58]: s4.str.contains('A', na=False)
Out[58]: 
0     True
1    False
2    False
3     True
4    False
5    False
6     True
7    False
8    False
dtype: bool

Creating Indicator Variables

You can extract dummy variables from string columns. For example if they are separated by a '|':

In [59]: s = pd.Series(['a', 'a|b', np.nan, 'a|c'])

In [60]: s.str.get_dummies(sep='|')
Out[60]: 
   a  b  c
0  1  0  0
1  1  1  0
2  0  0  0
3  1  0  1

See also get_dummies().

Method Summary

Method Description
cat() Concatenate strings
split() Split strings on delimiter
rsplit() Split strings on delimiter working from the end of the string
get() Index into each element (retrieve i-th element)
join() Join strings in each element of the Series with passed separator
get_dummies() Split strings on the delimiter returning DataFrame of dummy variables
contains() Return boolean array if each string contains pattern/regex
replace() Replace occurrences of pattern/regex with some other string
repeat() Duplicate values (s.str.repeat(3) equivalent to x * 3)
pad() Add whitespace to left, right, or both sides of strings
center() Equivalent to str.center
ljust() Equivalent to str.ljust
rjust() Equivalent to str.rjust
zfill() Equivalent to str.zfill
wrap() Split long strings into lines with length less than a given width
slice() Slice each string in the Series
slice_replace() Replace slice in each string with passed value
count() Count occurrences of pattern
startswith() Equivalent to str.startswith(pat) for each element
endswith() Equivalent to str.endswith(pat) for each element
findall() Compute list of all occurrences of pattern/regex for each string
match() Call re.match on each element, returning matched groups as list
extract() Call re.search on each element, returning DataFrame with one row for each element and one column for each regex capture group
extractall() Call re.findall on each element, returning DataFrame with one row for each match and one column for each regex capture group
len() Compute string lengths
strip() Equivalent to str.strip
rstrip() Equivalent to str.rstrip
lstrip() Equivalent to str.lstrip
partition() Equivalent to str.partition
rpartition() Equivalent to str.rpartition
lower() Equivalent to str.lower
upper() Equivalent to str.upper
find() Equivalent to str.find
rfind() Equivalent to str.rfind
index() Equivalent to str.index
rindex() Equivalent to str.rindex
capitalize() Equivalent to str.capitalize
swapcase() Equivalent to str.swapcase
normalize() Return Unicode normal form. Equivalent to unicodedata.normalize
translate() Equivalent to str.translate
isalnum() Equivalent to str.isalnum
isalpha() Equivalent to str.isalpha
isdigit() Equivalent to str.isdigit
isspace() Equivalent to str.isspace
islower() Equivalent to str.islower
isupper() Equivalent to str.isupper
istitle() Equivalent to str.istitle
isnumeric() Equivalent to str.isnumeric
isdecimal() Equivalent to str.isdecimal