Converting character values to numeric values in R with a function

Question

I'm looking to load and process a CSV file with seven variables, one which is a grouping variable / factor (data$hashtag) and six which are categories (data$support and others) denoted with either an "X" or "x" (or were left blank).

data <- read.csv("maet_coded_tweets.csv", stringsAsFactors = F)

names(data) <- c("hashtag", "support", "contributeConversation", "otherCommunities", "buildCommunity", "engageConversation", "unclear")

str(data)

'data.frame':   854 obs. of  7 variables:
 $ hashtag               : chr  "#capstoneisfun" "#capstoneisfun" "#capstoneisfun" "#capstoneisfun" ...
 $ support               : chr  "x" "x" "x" "x" ...
 $ contributeConversation: chr  "" "" "" "" ...
 $ otherCommunities      : chr  "" "" "" "" ...
 $ buildCommunity        : chr  "" "" "" "" ...
 $ engageConversation    : chr  "" "" "" "" ...
 $ unclear               : chr  "" "" "" "" ...

When I use a function to recode "X" or "x" to 1, and "" (blank) 0, the data are strangely character type, not numeric as intended.

recode <- function(x) {

  x[x=="x"] <- 1
  x[x=="X"] <- 1
  x[x==""] <- 0
  x
}

data[] <- lapply(data, recode)

str(data)

'data.frame':   854 obs. of  7 variables:
 $ hashtag               : chr  "#capstoneisfun" "#capstoneisfun" "#capstoneisfun" "#capstoneisfun" ...
 $ support               : chr  "1" "1" "1" "1" ...
 $ contributeConversation: chr  "0" "0" "0" "0" ...
 $ otherCommunities      : chr  "0" "0" "0" "0" ...
 $ buildCommunity        : chr  "0" "0" "0" "0" ...
 $ engageConversation    : chr  "0" "0" "0" "0" ...
 $ unclear               : chr  "0" "0" "0" "0" ...

When I tried to coerce the characters using as.numeric() in the function, it still didn't work. What gives - why would the variables be treated as characters and how to character variables to numeric?

vectors can only hold one datatype. Therefore, if you replace parts of a characterstring with a numeric variable it gets converted to character. How exactly did you use as.numeric() in the function? — Heroka
– Heroka, Commented Nov 12, 2015 at 13:39
recode <- function(x) { x[x=="x"] <- as.numeric(1) x[x=="X"] <- as.numeric(1) x[x==""] <- as.numeric(0) x } — Joshua Rosenberg
– Joshua Rosenberg, Commented Nov 12, 2015 at 13:41
You could try return(as.numeric(x)). As I said in my previous comment, the way you did that still forces conversion to character. Or you could do res <- ifelse(x %in% c("x","X"),1,0) — Heroka
– Heroka, Commented Nov 12, 2015 at 13:41

C8H10N4O2 · Accepted Answer · 2015-11-12 13:47:31Z

2

How about:

recode <- function(x) {
  ifelse(x %in% c('X','x'), 1,0)
}

Explanation: the steps in the function are evaluated sequentially, not simultaneously. So when you partially assign 1's to a character vector, these get converted to "1"s.

edited Nov 12, 2015 at 13:47

answered Nov 12, 2015 at 13:42

C8H10N4O2

19.2k10 gold badges106 silver badges145 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

C8H10N4O2 Over a year ago

For further practice, you can also convert logical to 0/1 with as.numeric(). So you could do recode2 <- function(x) as.numeric(x %in% c('X','x')) or even recode3 <- function(x) as.numeric(grepl('^x$',x, ignore.case=T))

talat Over a year ago

I'd add as.integer(tolower(x) == "x")

C8H10N4O2 Over a year ago

@docendodiscimus true, just a note that OP seemed to want numeric. however, a bool seems more logical :) than an int or float

sam · Accepted Answer · 2015-11-12 13:42:36Z

1

Waht about something like this?

# sample data with support being a character vector
data.frame(support = c("X","X","0","x","0"),a=1:5,stringsAsFactors = F)->myDat
# convert to a factor and check the order of the levels
myDat$support <- as.factor(myDat$support)
levels(myDat$support)
#"0" "x" "X"
# just to see that it worked make an additional variable
myDat$supportrecoded <- myDat$support
# change levels and convert
levels(myDat$supportrecoded) <- c("0","1","1")
myDat$supportrecoded <- as.integer(as.character(myDat$supportrecoded ))

answered Nov 12, 2015 at 13:42

sam

1587 bronze badges

Comments

narendra-choudhary · Accepted Answer · 2015-11-12 14:01:06Z

1

Using mapvalues from plyr.

data$support <- as.numeric(mapvalues(data$support, c("X", "x", ""), c(1, 1, 0)))

Using replace.

data$support <- replace(x <- data$support, x == "X", 1)
data$support <- replace(x <- data$support, x == "x", 1)
data$support <- replace(x <- data$support, x == "", 0)
data$support <- numeric(data$support)

answered Nov 12, 2015 at 14:01

narendra-choudhary

4,8664 gold badges43 silver badges62 bronze badges

Collectives™ on Stack Overflow

Converting character values to numeric values in R with a function

3 Answers 3

3 Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

3 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related