case_when()语句中新变量的变异因子

Question

数据设置

我有一个数据集，看起来有点像下面这个简单的dataframe：

CAD_EXCHANGE <- 1.34
EUR_EXCHANGE <- 0.88
 
df <- tibble(
  shipment = c("A", "B", "C", "D", "E"),
  invoice = c(rep(500, 5)),
  currency = factor(c("USD", "EUR", "CAD", NA, "SDD"))
)
 
df
# A tibble: 5 x 3
  shipment invoice currency
  <chr>      <dbl> <fct>   
1 A            500 USD     
2 B            500 EUR     
3 C            500 CAD     
4 D            500 NA      
5 E            500 SDD     

levels(df$currency)
[1] "CAD" "EUR" "SDD" "USD"

最终目标

我正试图将发票转换成美元的一些共同的其他货币(欧元和加元)，但不是全部，或如果数据丢失(即SDD和NA)。我的最终数据框架应该如下所示：

# A tibble: 5 x 5
  shipment invoice currency invoice_converted currency_converted
  <chr>      <dbl> <fct>                <dbl> <fct>             
1 A            500 USD                    500 USD               
2 B            500 EUR                    568 USD               
3 C            500 CAD                    373 USD               
4 D            500 NA                     500 NA                
5 E            500 SDD                    500 SDD               

试验1-不起作用

在未来，我可能要转换的不仅仅是这几种货币，所以我应用了一个case_when()语句。这是我第一次尝试：

df_USD1 <- df %>%
  mutate(
    invoice_converted = case_when(
      currency == "EUR" ~ round(invoice / EUR_EXCHANGE),
      currency == "CAD" ~ round(invoice / CAD_EXCHANGE),
      TRUE ~ invoice
    ),
    currency_converted = case_when(currency == "EUR" ~ "USD",
                                   currency == "CAD" ~ "USD",
                                   TRUE ~ currency)
  )

Error: Problem with `mutate()` column `currency_converted`.
i `currency_converted = case_when(...)`.
x must be a character vector, not a `factor` object.

有了以上这些，我了解到我在分配给currency_converted的任务中混合了字符和因素，因为我有默认的TRUE ~ currency ( currency是一个因素)。所以我试着用一些因素来完成任务.

试验2-有效，但不可靠

df_USD2 <- df %>%
  mutate(
    invoice_converted = case_when(
      currency == "EUR" ~ round(invoice / EUR_EXCHANGE),
      currency == "CAD" ~ round(invoice / CAD_EXCHANGE),
      TRUE ~ invoice
    ),
    currency_converted = case_when(
      currency == "EUR" ~ currency[1],
      currency == "CAD" ~ currency[1],
      TRUE ~ currency)
  )

这是可行的，但只是因为在我为这个问题的设置中，美元处于第一位，我不能依赖它。

> df$currency
[1] USD  EUR  CAD  <NA> SDD 
Levels: CAD EUR SDD USD

第三次试验--不起作用

我想我可以尝试一些其他的方法来获得这个因素，但这是行不通的：

df_USD3 <- df %>%
  mutate(
    invoice_converted = case_when(
      currency == "EUR" ~ round(invoice / EUR_EXCHANGE),
      currency == "CAD" ~ round(invoice / CAD_EXCHANGE),
      TRUE ~ invoice
    ),
    currency_converted = case_when(
      currency == "EUR" ~ df$currency[df$currency == "USD"],
      currency == "CAD" ~ df$currency[df$currency == "USD"],
      TRUE ~ currency
    )
  )

Error: Problem with `mutate()` column `currency_converted`.
i `currency_converted = factor(...)`.
x `currency == "EUR" ~ df$currency[df$currency == "USD"]`, `currency == "CAD" ~ df$currency[df$currency == "USD"]` must be length 5 or one, not 2.
Run `rlang::last_error()` to see where the error occurred.

这似乎是因为NA被退回.

> df$currency[df$currency == "USD"]
[1] USD  <NA>
Levels: CAD EUR SDD USD

...because，如果我回到原来的df，用其他货币代替NA，它就能工作了--但很明显，我需要能够将NA保存在属于它的地方。

我觉得有一些很好的方法可以做到这一点，但我错过了它，尽管阅读了各种因素，尝试了不同的事情。帮助?

Answer 1

case_when不自动进行类型转换--即currency是factor，而case_when中其他条件的返回只是character。因此，我们可以强制将currency转换为character，使所有的返回都是相同的类，并且它应该可以工作。

library(dplyr)
df %>%
  mutate(
    invoice_converted = case_when(
      currency == "EUR" ~ round(invoice / EUR_EXCHANGE),
      currency == "CAD" ~ round(invoice / CAD_EXCHANGE),
      TRUE ~ invoice
    ), currency_converted = case_when(currency == "EUR" ~ "USD",
                                   currency == "CAD" ~ "USD",
                                   TRUE ~ as.character(currency)))

-output

# A tibble: 5 × 5
  shipment invoice currency invoice_converted currency_converted
  <chr>      <dbl> <fct>                <dbl> <chr>             
1 A            500 USD                    500 USD               
2 B            500 EUR                    568 USD               
3 C            500 CAD                    373 USD               
4 D            500 <NA>                   500 <NA>              
5 E            500 SDD                    500 SDD             

如果我们想将它保持为factor，可以在case_when之后用factor包装，或者直接使用fct_recode而不是case_when

library(forcats)
df %>%
  mutate(
    invoice_converted = case_when(
      currency == "EUR" ~ round(invoice / EUR_EXCHANGE),
      currency == "CAD" ~ round(invoice / CAD_EXCHANGE),
      TRUE ~ invoice
    ), currency_converted = fct_recode(currency, USD = "EUR", USD = "CAD"))

-output

# A tibble: 5 × 5
  shipment invoice currency invoice_converted currency_converted
  <chr>      <dbl> <fct>                <dbl> <fct>             
1 A            500 USD                    500 USD               
2 B            500 EUR                    568 USD               
3 C            500 CAD                    373 USD               
4 D            500 <NA>                   500 <NA>              
5 E            500 SDD                    500 SDD