Manipulating Values ​​in Pandas via Regex

Question

This is actually the next question here . I was not clear in my previous question, and since he received the answer, I felt it was better to put a new question instead.

I have a dataframe as shown below:

Column1    Column2    Column3    Column4                     Column5
5FQ        1.047      S$55.3     UG44.2 as of 02/Jun/2016    S$8.2 mm
600        (1.047)    S$23.3     AG5.6 as of 02/Jun/2016     S$58 mm
KI2        1.695      S$5.35     RR59.5 as of 02/Jun/2016    S$705 mm
88G        0.0025     S$(5.3)    NW44.2 as of 02/Jun/2016    S$112 mm
60G        5.63       S$78.4     UG21.2 as of 02/Jun/2016    S$6.21 mm
90F        (5.562)    S$(88.3)   IG46.2 as of 02/Jun/2016    S$8 mm

I am trying to use regexto delete all words and letters, keeping only numbers. However, if the number is within (), I would like to make a negative number instead.

Desired Conclusion

Column1    Column2    Column3    Column4       Column5
5          1.047      55.3       44.2          8.2
600        -1.047     23.3       5.6           58
2          1.695      5.35       59.5          705
88         0.0025     -5.3       44.2          112
60         5.63       78.4       21.2          6.21
90         -5.562     -88.3      46.2          8

Is it possible? I tried playing with this code, but did not know what the appropriate combination should be regex.

df.apply(lambda x: x.astype(str).str.extract(r'(\d+\.?\d*)', expand=True).astype(np.float))

Answer 1

: $1,005A → 1005 ( 1- , Column3)

In [131]: df
Out[131]:
  Column1  Column2   Column3                   Column4    Column5
0     5FQ    1.047   $1,005A  UG44.2 as of 02/Jun/2016   S$8.2 mm
1     600  (1.047)    S$23.3   AG5.6 as of 02/Jun/2016    S$58 mm
2     KI2    1.695    S$5.35  RR59.5 as of 02/Jun/2016   S$705 mm
3     88G   0.0025   S$(5.3)  NW44.2 as of 02/Jun/2016   S$112 mm
4     60G     5.63    S$78.4  UG21.2 as of 02/Jun/2016  S$6.21 mm
5     90F  (5.562)  S$(88.3)  IG46.2 as of 02/Jun/2016     S$8 mm

In [132]: to_replace = [r'\(([\d\.]+)\)', r'.*?([\d\.\,\-]+).*', ',']

In [133]: vals = [r'-\1', r'\1', '']

In [134]: df.replace(to_replace=to_replace, value=vals, regex=True)
Out[134]:
  Column1 Column2 Column3 Column4 Column5
0       5   1.047    1005    44.2     8.2
1     600  -1.047    23.3     5.6      58
2       2   1.695    5.35    59.5     705
3      88  0.0025    -5.3    44.2     112
4      60    5.63    78.4    21.2    6.21
5      90  -5.562   -88.3    46.2       8

OLD answer:

, DataFrame.replace():

In [28]: to_replace = [r'\(([\d\.]+)\)', r'.*?([\d\.-]+).*']

In [29]: vals = [r'-\1', r'\1']

In [30]: df.replace(to_replace=to_replace, value=vals, regex=True)
Out[30]:
  Column1 Column2 Column3 Column4 Column5
0       5   1.047    55.3    44.2     8.2
1     600  -1.047    23.3     5.6      58
2       2   1.695    5.35    59.5     705
3      88  0.0025    -5.3    44.2     112
4      60    5.63    78.4    21.2    6.21
5      90  -5.562   -88.3    46.2       8

Answer 2

r1 = r'\((\d+\.?\d*)\)'
r2 = r'(-?\d+\.?\d*)'
df.stack().str.replace(r1, r'-\1', 1) \
          .str.extract(r2, expand=False).unstack()

Answer 3

You can come up with:

import re

def onlynumbers(value):
    if value.startswith('('):
        return '-' + value
    rx = re.compile(r'\d+[\d.]*')
    try:
        return rx.search(value).group(0)
    except:
        return value

df.applymap(onlynumbers)

This returns: