Select 在分组数据中操作的前 X 个观察值

Question

我观察了几天不同客户的购买数量（1 = 购买，0 = 不购买）。现在对于每一天，我想总结前两天和当天特定客户的购买数量，所以总共3天。

示例数据：

da <- data.frame(customer_id = c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3,4,4,4,4,4),
                 day = c("2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15"),
                 purchase = c(1,1,1,0,1,0,0,1,1,0,1,1,1,1,0,1,0,1,0,1))

> da
   customer_id        day purchase
1            1 2016-04-11        1
2            1 2016-04-12        1
3            1 2016-04-13        1
4            1 2016-04-14        0
5            1 2016-04-15        1
6            2 2016-04-11        0
7            2 2016-04-12        0
8            2 2016-04-13        1
9            2 2016-04-14        1
10           2 2016-04-15        0
11           3 2016-04-11        1
12           3 2016-04-12        1
13           3 2016-04-13        1
14           3 2016-04-14        1
15           3 2016-04-15        0
16           4 2016-04-11        1
17           4 2016-04-12        0
18           4 2016-04-13        1
19           4 2016-04-14        0
20           4 2016-04-15        1

我正在寻找的输出：

output_da <- data.frame(customer_id = c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3,4,4,4,4,4),
                 day = c("2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15","2016-04-11","2016-04-12","2016-04-13","2016-04-14","2016-04-15"),
                 purchase = c(1,1,1,0,1,0,0,1,1,0,1,1,1,1,0,1,0,1,0,1),
                 purchases_last_3_days = c(1,2,3,2,2,0,0,1,2,2,1,2,3,3,2,1,1,2,1,2))

output_da
   customer_id        day purchase purchases_last_3_days
1            1 2016-04-11        1                     1
2            1 2016-04-12        1                     2
3            1 2016-04-13        1                     3
4            1 2016-04-14        0                     2
5            1 2016-04-15        1                     2
6            2 2016-04-11        0                     0
7            2 2016-04-12        0                     0
8            2 2016-04-13        1                     1
9            2 2016-04-14        1                     2
10           2 2016-04-15        0                     2
11           3 2016-04-11        1                     1
12           3 2016-04-12        1                     2
13           3 2016-04-13        1                     3
14           3 2016-04-14        1                     3
15           3 2016-04-15        0                     2
16           4 2016-04-11        1                     1
17           4 2016-04-12        0                     1
18           4 2016-04-13        1                     2
19           4 2016-04-14        0                     1
20           4 2016-04-15        1                     2

我知道 cumsum 函数，但我不知道如何 select 当前每一天之前的 X 行数。

Answer 1

如果您只需要最后 3 行，您可以手动完成：

library(dplyr)
output_da <- da %>% 
    group_by(customer_id) %>% 
    mutate(pday = lag(purchase, default = 0), 
           apday = lag(purchase, n = 2, default = 0), 
           purchases_last_3_days = rowSums(across(c(purchase, pday, apday))),
           pday = NULL, apday = NULL)

更新

如果您需要的不仅仅是最后 3 行，可以稍微自动化一点，将 3 更改为合适的数字：

library(dplyr)
library(purrr)
last_day_rows <- 3
da %>% 
    group_by(customer_id) %>% 
    mutate(purchases_last_3_days = 
           rowSums(map_dfc(c(1:last_day_rows), 
           ~lag(purchase, n = .x - 1, default = 0))))

Answer 2

这是一个基本的 R 选项，来自 ave

output_da <- within(
  da,
  purchases_last_3_days <- ave(purchase,
    customer_id,
    FUN = function(v) sapply(seq_along(v), function(k) sum(v[pmax(k - 2, 1):k]))
  )
)

Select 在分组数据中操作的前 X 个观察值

Select precending X observations for operation in grouped data

r

panel

cumsum

dplyr

data-cleaning