Splittig a column of a dataframe into two columns with multiple conditions-R

Question

Let say I have such a dataframe(df):

header
---------
23.43
1,23
34.54
56.2
5,5

Some numbers are with dot(.) and some are with comma(,).

I want to split this column into two columns according to dots and commas into a new dataframe(df2) like:

header1  header2
------- --------
23       43
1        23
34       54
56       2
5        5

How can I do this using R? I will be very glad for any help. Thanks a lot.

Thanks a lot for the answers and for your time.

What about if the separators are in more than one character. For example:

header
---------
23 ab 43
1 ca 23
34 ab 54
56 ca 2
5 ca 5

How can we split this dataframe into two columns according to seperators "ab" and "ca"?

Also, `library(data.table) ; setDT(df)[, tstrsplit(header, "[,.]", type.convert = TRUE)]` which will make the correct type conversion too. — David Arenburg, Mar 07 '16 at 21:08
@docendodiscimus `tidyr::separate(df, header, c("h1", "h2"))` will also work, as it has a default `sep = "[^[:alnum:]]+"` — David Arenburg, Mar 07 '16 at 21:17
@DavidArenburg, that's interesting, though I prefer to be more explicit. Also, this will allow splitting on other symbols than just `.,`, as you know, of course — talat, Mar 07 '16 at 21:20

sgibb · Accepted Answer · 2016-03-07T21:12:05.687

4

Based on @alistaire's comment you can combine strsplit (split the strings), do.call and rbind (bind results rowise):

df <- data.frame(header=c("23.43", "1,23", "34.54", "56.2", "5,5"),
                 stringsAsFactors = FALSE)


r <- as.data.frame(do.call(rbind, strsplit(df$header, "[,.]")),
                   stringsAsFactors = FALSE)
colnames(r) <- c("header1", "header2")
r[] <- lapply(r, type.convert)
r
#   header1 header2
# 1      23      43
# 2       1      23
# 3      34      54
# 4      56       2
# 5       5       5

edited Mar 07 '16 at 21:12

answered Mar 07 '16 at 21:01

sgibb

25,396
3
68
74

1

I think `lapply(r, type.convert)` is the safest way – David Arenburg Mar 07 '16 at 21:10
@DavidArenburg: edited. Why `type.convert` is safer than `as.numeric`? – sgibb Mar 07 '16 at 21:13
1

Because you don't need to know the column classes apriroy. OP may some times have, character, logical, numeric, etc. This will always work. – David Arenburg Mar 07 '16 at 21:14
@DavidArenburg: Ok, thanks! – sgibb Mar 07 '16 at 21:15
@DavidArenburg, `type.convert` will only work if you supply character vectors to it. So you probably need something like `lapply(r, function(x) type.convert(as.character(x)))`. See for example `lapply(iris, type.convert)` – talat Mar 11 '16 at 08:42
@docendodiscimus I know that. But the result of a `strsplit` is always a `character`. – David Arenburg Mar 11 '16 at 09:29

Splittig a column of a dataframe into two columns with multiple conditions-R

1 Answers1