Manipulate R Dataframe conditionally

Question

I have a dataframe with integers and I would like to convert them into a binary system (or tertiary if you will) where they become 1 if greater than x, -1 if smaller than y, and 0 else.

This is an example dataframe:

  var1 var2 var3
  30    13   2
  20    29   3

This is what my new dataframe should look like (x is 27 and y is 4):

  var1 var2 var3
  1     0   - 1
  0     1    -1

Is there a simple way of doing this?

Answer 1

Here is a pretty quick base R answer. This will be super fast as long as the data set is fairly small relative to the amount of available RAM.

dat[] <- findInterval(as.matrix(dat), vec = c(4, 27),
                      rightmost.closed=TRUE) - 1L

Here, since each column has the same breaks, then you convert a copy of the data.frame to a matrix and run findInterval on using those breaks. The rightmost.closed=TRUE makes sure that these values are included in their set. Then, since findInterval returns values beginning with 0, subtract 1 to get the desired values.

Using dat[] <- puts the resulting vector into the data.frame.

This returns

dat
  var1 var2 var3
1    1    0   -1
2    0    1   -1

data

dat <- 
structure(list(var1 = c(30L, 20L), var2 = c(13L, 29L), var3 = 2:3),
.Names = c("var1", "var2", "var3"), class = "data.frame",
row.names = c(NA, -2L))

Answer 2

Here's a relatively succinct way to manage this with mutate_all and case_when from dplyr :

x <- 27
y <- 4
df %>% mutate_all(funs(case_when(. > x ~ 1, . < y ~ -1, TRUE ~ 0)))

#   var1 var2 var3
# 1    1    0   -1
# 2    0    1   -1

This can also be done with nested application of ifelse , though it's less extensible (ie pretty quickly becomes unwieldy if your list of conditions grows):

ifelse(df > x, 1, ifelse(df < y, -1, 0))

Though since you mention that you're doing "tertiary" encoding, perhaps that's all you need.