根据变量排名创建一个虚拟变量

Question

我有一个数据集，其中包含从 HHID 到 hum_aid 的列，并希望根据汇款、govt_aid 和 hum_aid 列排名前 2 的条件创建列

desired

，即前 2 列应至少是 3 列之一，如以下数据帧所示：

df <- structure(list(HHID = c(1, 2, 3, 4, 5, 6, 7, 8, 9), empl = c(1000, 
1750, 480, 630, 1200, 1000, 1700, 500, 300), agric = c(750, 650, 
400, 400, 0, 750, 0, 0, 750), remittances = c(0, 200, 1500, 1250, 
0, 500, 700, 900, 1200), govt_aid = c(0, 0, 1200, 1000, 0, 0, 
0, 900, 1300), loans = c(0, 350, 300, 0, 0, 1200, 500, 250, 500
), assets = c(500, 0, 0, 0, 700, 0, 0, 500, 150), hum_aid = c(0, 
0, 800, 1200, 0, 0, NA, 1400, 1500), desired = c(0, 0, 1, 1, 
0, 0, 0, 1, 1)), class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, 
-9L))

PS：我希望有一个不涉及重塑的解决方案。 预先感谢！

Answer 1

这是一个相对简单的方法：

df |>
  # for each row,
  rowwise() |>
  # check if remittances, government aid or humanitarian aid is equal to the top two values. If in more than one column, the answer is yes, then TRUE
  mutate(d = sum(map_lgl(c(remittances, govt_aid, hum_aid), ~ .x %in% tail(sort(c_across(empl:hum_aid)), 2))) > 1) |>
  ungroup()

输出：

# A tibble: 9 × 10
   HHID  empl agric remittances govt_aid loans assets hum_aid desired d    
  <dbl> <dbl> <dbl>       <dbl>    <dbl> <dbl>  <dbl>   <dbl>   <dbl> <lgl>
1     1  1000   750           0        0     0    500       0       0 FALSE
2     2  1750   650         200        0   350      0       0       0 FALSE
3     3   480   400        1500     1200   300      0     800       1 TRUE 
4     4   630   400        1250     1000     0      0    1200       1 TRUE 
5     5  1200     0           0        0     0    700       0       0 FALSE
6     6  1000   750         500        0  1200      0       0       0 FALSE
7     7  1700     0         700        0   500      0      NA       0 FALSE
8     8   500     0         900      900   250    500    1400       1 TRUE 
9     9   300   750        1200     1300   500    150    1500       1 TRUE 

Answer 2

一次性计算所有行的解决方案。我假设您想要

desired=1

如果提到的列中的 any 排名为 1-2。

library(dplyr)
ranks <- df |>
  transmute(across(c(remittances, govt_aid, hum_aid),
                   ~ rank(-.x, ties.method = "min") <= 2)) |>
  rowSums()
df |>
  mutate(desired2 = +(ranks > 0))
# # A tibble: 9 × 10
#    HHID  empl agric remittances govt_aid loans assets hum_aid desired desired2
#   <dbl> <dbl> <dbl>       <dbl>    <dbl> <dbl>  <dbl>   <dbl>   <dbl>    <int>
# 1     1  1000   750           0        0     0    500       0       0        0
# 2     2  1750   650         200        0   350      0       0       0        0
# 3     3   480   400        1500     1200   300      0     800       1        1
# 4     4   630   400        1250     1000     0      0    1200       1        1
# 5     5  1200     0           0        0     0    700       0       0        0
# 6     6  1000   750         500        0  1200      0       0       0        0
# 7     7  1700     0         700        0   500      0      NA       0        0
# 8     8   500     0         900      900   250    500    1400       1        1
# 9     9   300   750        1200     1300   500    150    1500       1        1