用R中的NA替换文本

Question

I have a variable PC_R in dataframe jd_df that describes laboratory test results. I'd like to replace some of the data in this variable (eg tf, QNS, rej) with NA. I've tried this code:

 jd_df %>%
  replace(PC_R,TF,NA)

and this:

jd_df %>%
  replace(jd_df,PC_R==TF,NA)    

and this:

jd_df %>%
  replace(PC_R,"TF","NA")

and this:

jd_df %>%
  replace(jd_df,PC_R%in%TF,NA)

I keep getting the error:

Error in replace(., jd_df, PC_R %in% TF, NA) : unused argument (NA)

I'm wondering if the replace command is not the way to go.

Answer 1

You can achieve this using case_when as demonstrated using the iris dataset below

library(dplyr)
iris <- iris %>% 
  mutate(Species = as.character(Species)) %>% 
  mutate(Species = case_when(
    Species == "setosa" ~ NA_character_, 
    TRUE ~ Species
  ))

Multiple changes can be specified as such:

iris %>% 
  mutate(Species = as.character(Species)) %>% 
  mutate(Species = case_when(
    Species == "setosa" | Species == "versicolor" ~ NA_character_, 
    TRUE ~ Species
  ))

Created on 2019-02-16 by the reprex package (v0.2.0).

Answer 2

我发现这可以将多个字符文本更改为NA：

jd_df %>% mutate(PC_R = replace(PC_R, PC_R %in% "TF"|PC_R %in% "tf"|PC_R %in% "rej",NA))

Answer 3

The case_when() answer above works well! A simpler alternative is the na_if() function, which replaces a specified string with NA . As such:

library(dplyr)
iris %>%
  mutate(Species = na_if(Species, "setosa")) 

This will change all instances of setosa to NA in the column Species . In your case, it could be:

jd_df %>%
  mutate(PC_R = na_if(PC_R, "TF"))

Which replaces all "TF" with NA . You can repeat the code as needed to catch all your intended NA values:

jd_df %>%
  mutate(PC_R = na_if(PC_R, "TF"),
         PC_R = na_if(PC_R, "QNS"),
         PC_R = na_if(PC_R, "rej"))