如何以整洁的方式重新编码众多因子变量

Question

我有许多变量，这些变量本质上是我想重新编码为整数的因素。

多个变量是一个字符串，第一个字符是对应于整数例如2 = I have considered suicide in the past week, but not made any plans. 应该是 2。其他变量是 yes 或 no，应该分别是 1 或 0。其他的，有许多基于多个字符串的级别：

none = 0
one = 1
two = 2
three = 3
four or more = 4

同样：

ptsd = 0
depression = 1
generalised anxiety = 2
no diagnosis warranted = 3

并且：

Female = 0
Male = 1
Other = 2

单元格中的一些值是 NA，需要保持 NA。我试过尝试以下代码而不尝试更改所有变量（从简单开始）：

vars1 <- vars(pastpsyc, pastmed, hxsuicide)
vars2 <- vars(siss, mssi_1)

df_rc <- df %>%
    ## this works
    mutate_at(vars1, ~ (case_when(
        . == "yes" ~ 1,
        . == "no" ~ 0
    ))) %>%
    ## this does not
    mutate_at(vars2, ~as.integer(str_extract(vars2, "[0-9]"))) %>%
    ## nor does this
    mutate_at(diag1, ~ (case_when(
        . == "ptsd" ~ 0,
        . == "depression" ~ 1,
        . == "generalised anxiety" ~ 2,
        . == "no diagnosis warranted" ~ 3
    )))

但这失败了，我完全不知道如何重新编码其他变量。

如何将不同的字符串更改为我需要的格式（最好以整洁的方式）？下面是一个最小可重现的数据集。

structure(list(siss = c("2 = I have considered suicide in the past week, but not made any plans.",
"1 = I have had vague thoughts of suicide in the past week.",
"2 = I have considered suicide in the past week, but not made any plans.",
"1 = I have had vague thoughts of suicide in the past week.",
"3 = I have made plans to suicide in the past week, but I haven’t intended to act on these plans."
), mssi_1 = c("1. Weak - unsure about whether he/she wants to die, seldom thinks about death, or intensity seems low.",
"1. Weak - unsure about whether he/she wants to die, seldom thinks about death, or intensity seems low.",
"1. Weak - unsure about whether he/she wants to die, seldom thinks about death, or intensity seems low.",
"1. Weak - unsure about whether he/she wants to die, seldom thinks about death, or intensity seems low.",
"2. Moderate - current desire to die, may be preoccupied with ideas about death, or intensity seems greater than a rating of 1."
), diag1 = c("ptsd", NA, "depression", "generalised anxiety",
"no diagnosis warranted"), pastpsyc = c("yes", NA, "no", NA,
"yes"), pastmed = c("no", "yes", NA, "no", "no"), hxsuicide = c("yes",
NA, "yes", "yes", "yes"), suicide_attempts = c("none", NA, "one",
"two", "four or more"), sex = c("Male", "Other", NA, "Female",
NA)), class = c("spec_tbl_df", "tbl_df", "tbl", "data.frame"), row.names = c(NA,
-5L), spec = structure(list(cols = list(siss = structure(list(), class = c("collector_character",
"collector")), mssi_1 = structure(list(), class = c("collector_character",
"collector")), diag1 = structure(list(), class = c("collector_character",
"collector")), pastpsyc = structure(list(), class = c("collector_character",
"collector")), pastmed = structure(list(), class = c("collector_character",
"collector")), hxsuicide = structure(list(), class = c("collector_character",
"collector")), suicide_attempts = structure(list(), class = c("collector_character",
"collector")), sex = structure(list(), class = c("collector_character",
"collector"))), default = structure(list(), class = c("collector_guess",
"collector")), skip = 1), class = "col_spec"))

Answer 1

这只需要很小的改动。

您想要在 str_extract

vars2

您希望 vars(diag1) 或 "diag1"（或仅使用 mutate）更改该单列

希望这对您有所帮助。

df %>%
  ## this works
  mutate_at(vars1, ~ (case_when(
    . == "yes" ~ 1,
    . == "no" ~ 0
  ))) %>%
  ## this does not
  mutate_at(vars2, ~as.integer(str_extract(., "[0-9]"))) %>%
  ## nor does this
  mutate_at(vars(diag1), ~ (case_when(
    . == "ptsd" ~ 0,
    . == "depression" ~ 1,
    . == "generalised anxiety" ~ 2,
    . == "no diagnosis warranted" ~ 3
  )))

如果您想使用 mutate 而不是 mutate_at:

mutate(diag1 = case_when(
  diag1 == "ptsd" ~ 0,
  diag1 == "depression" ~ 1,
  diag1 == "generalised anxiety" ~ 2,
  diag1 == "no diagnosis warranted" ~ 3
))

如何以整洁的方式重新编码众多因子变量

How to recode numerous factor variables in a tidy manner

r

stringr

dplyr

recode