仅当 R (dplyr) 中满足特定条件时，才将一列中的某些数据替换为另一列数据

Question

我有一个包含 30 多列和 10000 多行的广泛数据框。今天我想重点关注两列：languages 和 languages2：

languages             languages2

Spanish                  NA
Spanish                  NA
Other (specify)        French
Other (specify)        German
Other (specify)        Russian
English                  NA
Other (specify)        Portuguese
English                  NA
(...)

这就是我所需要的：

languages

Spanish
Spanish
French
German
Russian
English
Portuguese
English
(...)

我正在使用 mutate 函数寻找答案dplyr

Answer 1

对于更大的数据，您可能需要控制其他场景，例如防止用 NA 替换数据并保留 Other 值。此外，如果解决第一列可能包含 Other() 或 Other, lang 的场景可能有好处。您可能需要考虑使用正则表达式或预处理第一列。

library("tidyverse")
dta <- tribble(
    ~lang1, ~lang2,
    "Spanish", NA,
    "Other", "English",
    "Other", NA
)

mutate(dta, lang1 = case_when(
    grepl("^Other,*", lang1) & !is.na(lang2) ~ lang2,
    TRUE ~ lang1
))

Answer 2

使用dplyr，我们可以将Other (specify)替换为NA，然后使用coalesce：

library(dplyr)

df %>%
  mutate(languages = coalesce(na_if(languages, "Other (specify)"), languages2)) %>%
  select(languages)

输出< /strong>

  languages
1    Spanish
2    Spanish
3     French
4     German
5    Russian
6    English
7 Portuguese
8    English

tidyverse 选项是使用 str_replace_all 将 Other (specify) 替换为 languages2 中的值。

library(tidyverse)

df %>%
  mutate(languages = str_replace_all(languages,"Other \\(specify\\)", languages2)) %>% 
  select(languages)

数据

df <- structure(list(languages = c("Spanish", "Spanish", "Other (specify)", 
"Other (specify)", "Other (specify)", "English", "Other (specify)", 
"English"), languages2 = c(NA, NA, "French", "German", "Russian", 
NA, "Portuguese", NA)), class = "data.frame", row.names = c(NA, 
-8L))

基准

但是，如果您有大量数据并且需要更快的速度，那么您可能会考虑基本 R，它比 dplyr 更快或data.table。

bm <- microbenchmark::microbenchmark(Konrad = mutate(df, languages = case_when(
  grepl("^Other,*", languages) & !is.na(languages2) ~ languages2,
  TRUE ~ languages
)),
langtang = df %>%
  mutate(languages=if_else(languages=="Other (specify)", languages2, languages)),
valentina = df %>%
  mutate(languages=if_else(!is.na(languages2), languages2, languages)),
andrew_stringr = df %>%
  mutate(languages = str_replace_all(languages,"Other \\(specify\\)", languages2)),
andrew_coalesce = df %>%
  mutate(languages = coalesce(na_if(languages, "Other (specify)"), languages2, 'Other (specify)')),
andrew_baseR = {df1 <- df; df1[df1$languages == "Other (specify)", "languages"] <- df1[df1$languages == "Other (specify)", "languages2" ]},
andrew_baseR_with = {df2 <- df; df2$languages <- with( df2, ifelse( languages == "Other (specify)", languages2, languages ) )},
andrew_datatable = {dt <- as.data.table(df); dt[languages == "Other (specify)", languages := languages2 ]},
times = 1000)

Answer 3

data %>% 
  mutate(languages=if_else(languages=="Other (specify)", languages2, languages)

Answer 4

另一种可能性（如果您希望检查位于第二列而不是第一列）：

library(dplyr)

df <- data.frame(languages=c("Spanish","Spanish","Other (specify)","Other (specify)","Other (specify)","English","Other (specify)","English"),languages2=c(NA,NA,"French","German","Russian",NA,"Portuguese",NA))

df %>%
 mutate(languages=if_else(!is.na(languages2), languages2, languages))

   languages languages2
1    Spanish       <NA>
2    Spanish       <NA>
3     French     French
4     German     German
5    Russian    Russian
6    English       <NA>
7 Portuguese Portuguese
8    English       <NA>