如何同時使用ifelse和group_by？

我在下面有這個資料。我想用中位數的 10 倍來估算收入的大值。這是使用以下代碼完成的：

df$income_imputed = ifelse(df$income > (10* median(df$income,na.rm = T)), (10* median(df$income,na.rm = T)), df$income)

但是，我想分別針對每個國家和年份而不是整個資料集執行此操作。我知道group_by可能對此類任務有所幫助，但我不確定如何將這兩個功能結合在一起。

    country  year   income
     <dbl> <dbl>    <dbl>
 1       1  1999     5000
 2       1  1999     5000
 3       1  1999 10000000
 4       1  1999     3000
 5       1  2000     4000
 6       1  2000     4000
 7       1  2000 20000000
 8       1  2000     4000
 9       2  1999    10000
10       2  1999    10000
11       2  1999 30000000
12       2  1999     4000
13       2  2000    12000
14       2  2000    12000
15       2  2000 40000000


df= structure(list(country = c(1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 
2, 2, 2), year = c(1999, 1999, 1999, 1999, 2000, 2000, 2000, 
2000, 1999, 1999, 1999, 1999, 2000, 2000, 2000), income = c(5000, 
5000, 1e 07, 3000, 4000, 4000, 2e 07, 4000, 10000, 10000, 3e 07, 
4000, 12000, 12000, 4e 07), income2 = c(5000, 5000, 1e 05, 3000, 
4000, 4000, 1e 05, 4000, 10000, 10000, 1e 05, 4000, 12000, 12000, 
1e 05)), row.names = c(NA, -15L), class = c("tbl_df", "tbl", 
"data.frame"))

uj5u.com熱心網友回復：

我相信這是你所期望的：

my_df <- my_df %>% group_by(country, year) %>% mutate(income_imputed = ifelse(income > (10* median(income,na.rm = T)), (10* median(income,na.rm = T)), income))

uj5u.com熱心網友回復：

df  
   country  year   income income2
         <dbl> <dbl>    <dbl>   <dbl>
     1       1  1999     5000    5000
     2       1  1999     5000    5000
     3       1  1999 10000000  100000
     4       1  1999     3000    3000
     5       1  2000     4000    4000
     6       1  2000     4000    4000
     7       1  2000 20000000  100000
     8       1  2000     4000    4000
     9       2  1999    10000   10000
    10       2  1999    10000   10000
    11       2  1999 30000000  100000
    12       2  1999     4000    4000
    13       2  2000    12000   12000
    14       2  2000    12000   12000
    15       2  2000 40000000  100000

使用 dplyr：

  df %>% group_by(country,year) %>% mutate(income_imputed =ifelse(income > (10* median(income,na.rm = T)), (10* median(income,na.rm = T)),income))

輸出：

 country  year   income income2 income_imputed
     <dbl> <dbl>    <dbl>   <dbl>          <dbl>
 1       1  1999     5000    5000           5000
 2       1  1999     5000    5000           5000
 3       1  1999 10000000  100000          50000
 4       1  1999     3000    3000           3000
 5       1  2000     4000    4000           4000
 6       1  2000     4000    4000           4000
 7       1  2000 20000000  100000          40000
 8       1  2000     4000    4000           4000
 9       2  1999    10000   10000          10000
10       2  1999    10000   10000          10000
11       2  1999 30000000  100000         100000
12       2  1999     4000    4000           4000
13       2  2000    12000   12000          12000
14       2  2000    12000   12000          12000
15       2  2000 40000000  100000         120000

uj5u.com熱心網友回復：

將 group_by() 放在計算之前應該可以解決問題

library(dplyr)

df <- df %>% 
      group_by(country, year) %>% 
      mutate(income2 = ifelse(income > 10 * median(income, na.rm =  TRUE), 
                              yes = 10 * median(income, na.rm = TRUE), 
                              no = income))

標籤：r dplyr

上一篇：如何在年尺度上繪制月份值？

下一篇：在pivot_longer之前清理標題