Changing all values in a data frame

Often you will inherit a data frame in which there is a systematic mistake in data entry. Often this involves how absent, NA, or zero values have to be treated. We can make large scale changes to the dataframe using the square brackets [ ] in base R (see page 50 of book). Let’s make a small mock data frame:

x <- LETTERS[1:4]
y <- c(0,1,0,1)
z <- c(0,1,1,1)

original <- data.frame(x,y,z)

Let’s look at our orginal dataframe:

original

  x y z
1 A 0 0
2 B 1 1
3 C 0 1
4 D 1 1

WARNING: the following code will make permanent changes to your object. So as a consequence we will make copies based on our orginal data frame (i.e. v1, v2)

v1 <- original
v2 <- original

For example we could change all 1s to 2s. To do this we use the logical operators (given on page 113). In this example all elements of v1 which equal 1 will be changed to 2:

v1[v1==1]  <- 2

v1

Which results in:

  x y z
1 A 0 0
2 B 2 2
3 C 0 2
4 D 2 2

Using this method we could even change numbers to text:

v2[v2==1]  <- "cat"

v2

Which results in:

  x   y   z
1 A   0   0
2 B cat cat
3 C   0 cat
4 D cat cat

Now let’s change all zero values to NA. using the same procedure. In this example all elements of v2 which equal zero will be turned to NA:

v2[v2==0]  <- NA

v2

Which results in :

  x  y  z
  x    y    z
1 A <NA> <NA>
2 B  cat  cat
3 C <NA>  cat
4 D  cat  cat

However while changing numbers or text is straight forward, changing NAs back to some value (such as zero) requires a slightly different approach. Here we have to use the is.na() function as, unfortunately, v2[v2==NA] <- 0 doesn’t work:

v2[is.na(v2)]  <- 0 

v2

Which results in:

  x   y   z
1 A   0   0
2 B cat cat
3 C   0 cat
4 D cat cat

BACK to INDEX