Replace all particular values in a data frame

Having a data frame, how do I go about replacing all particular values along all rows and columns. Say for example I want to replace all empty records with NA's (without typing the positions):

df <- data.frame(list(A=c("", "xyz", "jkl"), B=c(12, "", 100))) A B
1 12
2 xyz
3 jkl 100

Expected result:

 A B
1 NA 12
2 xyz NA
3 jkl 100

5 Answers

Like this:

> df[df==""]<-NA
> df A B
1 <NA> 12
2 xyz <NA>
3 jkl 100

5

Since PikkuKatja and glallen asked for a more general solution and I cannot comment yet, I'll write an answer. You can combine statements as in:

> df[df=="" | df==12] <- NA
> df A B
1 <NA> <NA>
2 xyz <NA>
3 jkl 100

For factors, zxzak's code already yields factors:

> df <- data.frame(list(A=c("","xyz","jkl"), B=c(12,"",100)))
> str(df)
'data.frame': 3 obs. of 2 variables: $ A: Factor w/ 3 levels "","jkl","xyz": 1 3 2 $ B: Factor w/ 3 levels "","100","12": 3 1 2

If in trouble, I'd suggest to temporarily drop the factors.

df[] <- lapply(df, as.character)

Here are a couple dplyr options:

library(dplyr)
# all columns:
df %>% mutate_all(~na_if(., ''))
# specific column types:
df %>% mutate_if(is.factor, ~na_if(., ''))
# specific columns:
df %>% mutate_at(vars(A, B), ~na_if(., ''))
# or:
df %>% mutate(A = replace(A, A == '', NA))
# replace can be used if you want something other than NA:
df %>% mutate(A = as.character(A)) %>% mutate(A = replace(A, A == '', 'used to be empty'))

2

We can use data.table to get it quickly. First create df without factors,

df <- data.frame(list(A=c("","xyz","jkl"), B=c(12,"",100)), stringsAsFactors=F)

Now you can use

setDT(df)
for (jj in 1:ncol(df)) set(df, i = which(df[[jj]]==""), j = jj, v = NA)

and you can convert it back to a data.frame

setDF(df)

If you only want to use data.frame and keep factors it's more difficult, you need to work with

levels(df$value)[levels(df$value)==""] <- NA

where value is the name of every column. You need to insert it in a loop.

2

If you want to replace multiple values in a data frame, looping through all columns might help.

Say you want to replace "" and 100:

na_codes <- c(100, "")
for (i in seq_along(df)) { df[[i]][df[[i]] %in% na_codes] <- NA
}

Your Answer

Sign up or log in

Sign up using Google Sign up using Facebook Sign up using Email and Password Submit

Post as a guest

Post Your Answer Discard

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy