从数据帧信息中提取摘要

Question

data <- 
STUDY ID  BASE  CYCLE1   DIED  PROG
  1    1    100    30     No    Yes
  1    2    NA     20     Yes   No
  1    3    16     NA     Yes   Yes 
  1    4    15     10     Yes   Yes

我想对以下几点作一个总结：

1. 有多少受试者有基线值和CYCLE1值？
2. 在1个人中，有多少人已经死亡？
3. 在1人中，有多少人已经死亡或死亡？

答案：

1. 2-受试者(占50%) ==>受试者1和4
2. 1-主题(25%) ===>这是主题4
3. 2-科目(50%) ==>科目1和4

对此逐项研究的汇总表将是很棒的(显示数字和百分比)。我在使用Rstudio。

Answer 1

如果它是基于第一个filter

library(dplyr)
library(stringr)
data %>%
   group_by(STUDY) %>%
   filter(!is.na(BASE) & !is.na(CYCLE1)) %>%
   summarise(ID = str_c(ID, collapse=", "), 
             n1 = n(),
             n2 = sum(DIED== "Yes"), 
             n3 = sum(DIED == "Yes"|PROG == "Yes"))
# A tibble: 1 x 5
#  STUDY ID       n1    n2    n3
#  <int> <chr> <int> <int> <int>
#1     1 1, 4      2     1     2

如果我们也需要这个百分比

out <- data %>% 
        group_by(STUDY) %>%
        mutate(i1 = !is.na(BASE) & !is.na(CYCLE1),
          perc1 = 100 * mean(i1), 
          n1 = sum(i1), 
          i2 = DIED == "Yes" & i1, 
          perc2 = 100 * mean(i2),
          n2 = sum(i2), 
          i3 = (DIED == "Yes"|PROG == "Yes") & i1, 
          perc3 = 100 * mean(i3), 
          n3 = sum(i3)) %>%
        filter(i1) %>% 
        select(STUDY, ID, matches("perc"), matches("n")) %>% 
        mutate(ID = toString(ID)) %>% 
        slice(1)
# A tibble: 1 x 8
# Groups:   STUDY [1]
#  STUDY ID    perc1 perc2 perc3    n1    n2    n3
#  <int> <chr> <dbl> <dbl> <dbl> <int> <int> <int>
#1     1 1, 4     50    25    50     2     1     2

它可以进一步修改以格式化输出。

library(tidyr) # 0.8.3.9000
out %>% 
   pivot_longer(cols = perc1:n3, names_to = c( "perc", "n"), 
            names_sep = "(?<=[a-z])(?=[0-9])") %>%
   group_by(STUDY, ID, n) %>%
   summarise(value = sprintf("%d (%d%%)", last(value), first(value))) %>% 
   select(-n)