r语言 - 分析第一个值的多行依赖性?

Question

>我有这个表：

record_id result date_start   date_end
1          1    pos                      
2          1        26/06/2019 28/06/2019
3          1        27/06/2019 29/06/2019
4          1        28/06/2019 30/06/2019
5          1        29/06/2019 01/07/2019
6          2    neg                      
7          2        01/07/2019 03/07/2019
8          2        02/07/2019 04/07/2019
9          2        03/07/2019 05/07/2019
10         2        04/07/2019 06/07/2019
11         2        05/07/2019 07/07/2019
12         3    pos                      
13         3        07/07/2019 09/07/2019
14         3        08/07/2019 10/07/2019

我想计算每行的日期差异，没问题。之后，我想要的是分别分析"pos"和"neg"的组。但是当我有日期时，我的数据中没有结果的值。这是从 REDCap 导入的数据，带有重复仪器。 我使用tidyverse，我认为dplyr可以提供帮助，这不是我必须做的pivot_wider吗？我试过了，但没办法...

如果有人可以提供帮助，谢谢...

Answer 1

像这样，例如，计算每组的平均日期差异？

library(tidyverse)
library(lubridate)
df %>% 
fill(result, .direction = "down") %>% 
filter(!is.na(date_start)) %>% 
mutate(date_start = dmy(date_start),
date_end = dmy(date_end)) %>% 
group_by(result) %>% 
summarise(mean_date_dif = mean(date_end - date_start))
#`summarise()` ungrouping output (override with `.groups` argument)
## A tibble: 2 x 2
#  result mean_date_dif
#  <chr>  <drtn>       
#1 neg    2 days       
#2 pos    2 days 

---

数据

df <- tibble::tribble(
~record_id, ~result,  ~date_start,    ~date_end,
1L,   "pos",           NA,           NA,
1L,      NA, "26/06/2019", "28/06/2019",
1L,      NA, "27/06/2019", "29/06/2019",
1L,      NA, "28/06/2019", "30/06/2019",
1L,      NA, "29/06/2019", "01/07/2019",
2L,   "neg",           NA,           NA,
2L,      NA, "01/07/2019", "03/07/2019",
2L,      NA, "02/07/2019", "04/07/2019",
2L,      NA, "03/07/2019", "05/07/2019",
2L,      NA, "04/07/2019", "06/07/2019",
2L,      NA, "05/07/2019", "07/07/2019",
3L,   "pos",           NA,           NA,
3L,      NA, "07/07/2019", "09/07/2019",
3L,      NA, "08/07/2019", "10/07/2019"
)