我想找到一种有效的方法来总结一个团队在前 N 场比赛中获得的分数（不包括当前比赛中的分数），用于模拟 ID、球队和赛季的每个组合。 如果到目前为止，一支球队打的比赛少于 N 场，则该函数应返回 NA。

一个简化的数据集：

预期输出是数据中的一个额外列，称为“accumulated_team_points”，它返回该球队、赛季和模拟 ID 的最后 N 场比赛中的分数。

我似乎无法找到一种方法来做到这一点。

这里需要两个操作：分组（我将使用dplyr演示）和滚动窗口求和（我将使用zoo ）。

此外，您的数据有点过于稀疏，无法按team, season, simulation_ID的所有三个进行非常棒的分组，因此这次我将仅按team显示分组。 对于较大的数据，请将group_by(team)替换为group_by(team, season, simulation_ID) 。

library(dplyr)
k <- 3 # window size
dat %>%
  group_by(team) %>%
  mutate(accumulated_team_points = zoo::rollapply(team_points, k, FUN = sum, align = "right", fill = NA)) %>%
  ungroup()
# # A tibble: 10 x 8
#    match_ID season    simulation_ID home_team team           match_result team_points accumulated_team_points
#       <int> <chr>             <int> <lgl>     <chr>          <chr>              <int>                   <int>
#  1        1 2015-2016             1 TRUE      Manchester Utd Home win               3                      NA
#  2        1 2015-2016             2 TRUE      Manchester Utd Draw                   1                      NA
#  3        1 2015-2016             3 TRUE      Manchester Utd Home win               3                       7
#  4        1 2015-2016             1 FALSE     Tottenham      Home win               0                      NA
#  5        1 2015-2016             2 FALSE     Tottenham      Home win               0                      NA
#  6        1 2015-2016             3 FALSE     Tottenham      Away win               3                       3
#  7        2 2015-2016             1 TRUE      Leicester      Home win               3                      NA
#  8        2 2015-2016             2 TRUE      Leicester      Home win               3                      NA
#  9        2 2015-2016             3 TRUE      Leicester      Away win               0                       6
# 10        2 2015-2016             1 FALSE     Sunderland     Draw                   1                      NA     

对于窗口中的前k-1实例，默认为NA ，这通常是合理且可防御的默认值。 但是，如果您想对部分总和求和，则将fill=NA替换为partial=TRUE ：

dat %>%
  group_by(team) %>%
  mutate(accumulated_team_points = zoo::rollapply(team_points, k, FUN = sum, align = "right", partial = TRUE)) %>%
  ungroup()
# # A tibble: 10 x 8
#    match_ID season    simulation_ID home_team team           match_result team_points accumulated_team_points
#       <int> <chr>             <int> <lgl>     <chr>          <chr>              <int>                   <int>
#  1        1 2015-2016             1 TRUE      Manchester Utd Home win               3                       3
#  2        1 2015-2016             2 TRUE      Manchester Utd Draw                   1                       4
#  3        1 2015-2016             3 TRUE      Manchester Utd Home win               3                       7
#  4        1 2015-2016             1 FALSE     Tottenham      Home win               0                       0
#  5        1 2015-2016             2 FALSE     Tottenham      Home win               0                       0
#  6        1 2015-2016             3 FALSE     Tottenham      Away win               3                       3
#  7        2 2015-2016             1 TRUE      Leicester      Home win               3                       3
#  8        2 2015-2016             2 TRUE      Leicester      Home win               3                       6
#  9        2 2015-2016             3 TRUE      Leicester      Away win               0                       6
# 10        2 2015-2016             1 FALSE     Sunderland     Draw                   1                       1

仅供参考：我假设数据是预先订购的。

数据

dat <- structure(list(match_ID = c(1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L), season = c("2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016", "2015-2016"), simulation_ID = c(1L, 2L, 3L, 1L, 2L, 3L, 1L, 2L, 3L, 1L), home_team = c(TRUE, TRUE, TRUE, FALSE, FALSE, FALSE, TRUE, TRUE, TRUE, FALSE), team = c("Manchester Utd", "Manchester Utd", "Manchester Utd", "Tottenham", "Tottenham", "Tottenham", "Leicester", "Leicester", "Leicester", "Sunderland"), match_result = c("Home win", "Draw", "Home win", "Home win", "Home win", "Away win", "Home win", "Home win", "Away win", "Draw"), team_points = c(3L, 1L, 3L, 0L, 0L, 3L, 3L, 3L, 0L, 1L)), class = "data.frame", row.names = c(NA, -10L))