R - Replace specific value contents with NA [duplicate]

Answer 1

I believe the simplest solution is with base R function is.na<-. It's meant to solve precisely that issue.

First, make up some data. Then set the required values to NA.

set.seed(247)    # make the results reproducible

df <- data.frame(X = 1:10, Y = sample(c("-", letters[1:2]), 10, TRUE))

is.na(df) <- df == "-"
df
#    X    Y
#1   1    a
#2   2    b
#3   3    b
#4   4    a
#5   5 <NA>
#6   6    b
#7   7    a
#8   8 <NA>
#9   9    b
#10 10    a

Answer 2

Here's a solution that will do it:

> library(dplyr)
> test <- tibble(x = c('100', '20.56', '0.003', '-', '  -'),  y = 5:1)
> makeNA <- function(x) str_replace(x,'-',NA_character_)
> mutate_all(test, funs(makeNA))
# A tibble: 5 x 2
  x     y    
  <chr> <chr>
1 100   5    
2 20.56 4    
3 0.003 3    
4 NA    2    
5 NA    1  

Answer 3

Since you're already using tidyverse functions, you can easily use na_if from dplyr within your pipes.

For example, I have a dataset where 999 is used to fill in a non-answer:

df <- tibble(
    alpha = c("a", "b", "c", "d", "e"), 
    val1 = c(1, 999, 3, 8, 999), 
    val2 = c(2, 8, 999, 1, 2))

If I wanted to change val1 so 999 is NA, I could do:

df %>% 
    mutate(val1 = na_if(val1, 999))

In your case, it sounds like you want to replace a value across multiple variables, so using mutate_at or mutate_if would be more appropriate:

df %>%
    mutate_at(vars(val1, val2), na_if, 999)

replaces all instances of 999 in both val1 and val2 with NA and now looks like this:

# A tibble: 5 x 3
  alpha  val1  val2
  <chr> <dbl> <dbl>
1 a        1.    2.
2 b       NA     8.
3 c        3.   NA 
4 d        8.    1.
5 e       NA     2.