Trouble converting character to numeric in R

Question

I've pulled a table from Wikipedia, but I'm getting a bunch of junk with the population numbers I'm looking for. For instance, I get "!B9840748934017Â 8,244,910" when the actual number I'm after is 8,244,910 only. I've cleaned up the character vector with regex, using sub('![[:alnum:]]*[[:space:]]', '', x)

This works fine, leaving me with the character vector "8,244,910". When I try to convert it to numeric using as.numeric, however, it gets coerced to NA, and I'm unable to get an integer, no matter what conversions I try. Any thoughts?

I've added it as an answer. What page are you trying to get the data from? Just curious if there's a more direct way. — A5C1D2H2I1M1N2O1R2T1, Feb 05 '13 at 04:17
somewhat related question: [How to read a csv file in R where some numbers contain commas?](http://stackoverflow.com/q/1523126/59470) — topchef, Feb 05 '13 at 20:30

score 6 · Accepted Answer · answered Feb 05 '13 at 04:16

6

Try the following:

as.numeric(gsub('![[:alnum:]]*[[:space:]]|[[:punct:]]', '', x))

The problem is that you have commas in the output of your first attempt. Those need to be removed before converting to as.numeric.

answered Feb 05 '13 at 04:16

A5C1D2H2I1M1N2O1R2T1

190,393
28
405
485

score 3 · Answer 2 · answered Feb 05 '13 at 16:37

3

Ananda's solution does get the job done but something may go wrong:

[:punct:] also includes dot character (.) which is valid part of numeric;
bulky regex is hard to read and it's easier to break things up:

# remove junk

num_temp = sub('![[:alnum:]]*[[:space:]]', '', x)

# remove all commas from numerics

num = gsub(",", "", num_temp)

answered Feb 05 '13 at 16:37

topchef

19,091
9
63
102

+1, Good point about stripping decimals, but don't necessarily agree about having to break this problem into two steps.... – A5C1D2H2I1M1N2O1R2T1 Feb 06 '13 at 09:51

Trouble converting character to numeric in R

2 Answers2