使用 dplyr 和 mutate 创建基于组和最后 n 行的新列
Using dplyr and mutate to create new columns based on groups and last n rows
我以下面的数据框为例：
match_id <- c("match_1", "match_1","match_1","match_2","match_2","match_2","match_3","match_3","match_3", "match_4","match_4","match_4")
player_id <- c("player_1", "player_2", "player_3", "player_1", "player_2", "player_3", "player_1", "player_2", "player_3","player_1", "player_2", "player_3")
venue <- c("venue A", "venue A", "venue A", "venue B", "venue B", "venue B", "venue C", "venue C", "venue C","venue C", "venue C", "venue C")
opponent <- c("opponent A", "opponent A", "opponent A", "opponent B", "opponent B", "opponent B", "opponent C", "opponent C", "opponent C","opponent C", "opponent C", "opponent C")
points <- c(5,10,15,1,2,3,5,7,9,11,2,6)
data <- data.frame(match_id, player_id, venue, opponent, points)
我希望创建新的列，以根据分组显示点数列中的最后 n 个值。
第一组我的尝试，以3局为例：
#group by player, give the last n games
library(dplyr)
data <- data %>%
arrange(player_id,desc(match_id)) %>%
mutate(last3_games = lag(points, n=3)) %>%
mutate(last2_games = lag(points, n=2)) %>%
mutate(last1_games = lag(points, n=1))
给出：
head(data)
match_id player_id venue opponent points last3_games last2_games last1_games
1 match_4 player_1 venue C opponent C 11 NA NA NA
2 match_3 player_1 venue C opponent C 6 NA NA 11
3 match_2 player_1 venue B opponent B 1 NA 11 5
4 match_1 player_1 venue A opponent A 5 11 5 1
5 match_4 player_2 venue C opponent C 2 5 1 5
6 match_3 player_2 venue C opponent C 7 1 5 2
7 match_2 player_2 venue B opponent B 3 5 2 7
8 match_1 player_2 venue A opponent A 10 2 7 2
但我希望显示的值是每个玩家的最后 3 个点值，如下所示：
match_id player_id venue opponent points last3_games last2_games last1_games
1 match_4 player_1 venue C opponent C 11 1 6 11
2 match_3 player_1 venue C opponent C 6 5 1 6
3 match_2 player_1 venue B opponent B 1 NA 5 1
4 match_1 player_1 venue A opponent A 5 NA NA 5
5 match_4 player_2 venue C opponent C 2 3 7 2
6 match_3 player_2 venue C opponent C 7 10 3 7
7 match_2 player_2 venue B opponent B 3 NA 10 3
8 match_1 player_2 venue A opponent A 10 NA NA 10
然后我也想通过对手和场地做同样的事情：
#by opponent
match_id player_id venue opponent points last3_opponent last2_opponent last1_opponent
1 match_4 player_1 venue C opponent C 11 NA 5 11
2 match_3 player_1 venue C opponent C 5 NA NA 5
3 match_2 player_1 venue B opponent B 1 NA NA 1
4 match_1 player_1 venue A opponent A 5 NA NA 5
5 match_4 player_2 venue C opponent C 2 NA 7 2
6 match_3 player_2 venue C opponent C 7 NA NA 7
7 match_2 player_2 venue B opponent B 2 NA NA 2
8 match_1 player_2 venue A opponent A 10 NA NA 10
这里有很多NA's，但这是因为我只展示了一个小例子。
我的尝试是再次将它们分组并使用滞后进行变异，但它没有提供所需的输出：
data <- data %>%
arrange(player_id,desc(match_id), opponent) %>%
mutate(last3_opponent = lag(points, n=3)) %>%
mutate(last2_opponent = lag(points, n=2)) %>%
mutate(last1_opponent = lag(points, n=1))
按对手和场地分组的输出应如下所示：
#by opponent and venue
match_id player_id venue opponent points last3_opp_ven last2_opp_ven last1_opp_ven
1 match_1 player_1 venue A opponent A 5 NA NA 5
2 match_2 player_1 venue B opponent B 1 NA NA 1
3 match_4 player_1 venue C opponent C 11 NA 5 11
4 match_3 player_1 venue C opponent C 5 NA NA 5
5 match_1 player_2 venue A opponent A 10 NA NA 10
6 match_2 player_2 venue B opponent B 2 NA NA 2
7 match_4 player_2 venue C opponent C 2 NA 7 2
8 match_3 player_2 venue C opponent C 7 NA NA 7
理想情况下，我正在寻找最近 10 场比赛（最后 10 场、最后 9 场、最后 8 场等），并且由于有很多对手和很多 vanues，因此生成的数据框将有很多列。
有没有更简单的方法？
我还想在每个组的末尾添加一列，其中包含所有值的组合，对于每个组，就像这样（使用游戏的第一个示例）：
match_id player_id venue opponent points last3_combined
1 match_4 player_1 venue C opponent C 11 1,6,11
2 match_3 player_1 venue C opponent C 6 5,1,6
3 match_2 player_1 venue B opponent B 1 NA,5,1
4 match_1 player_1 venue A opponent A 5 NA,NA,5
5 match_4 player_2 venue C opponent C 2 3,7,2
6 match_3 player_2 venue C opponent C 7 10,3,7
7 match_2 player_2 venue B opponent B 3 NA,10,3
8 match_1 player_2 venue A opponent A 10 NA,NA,10
更新：
假设我想使用另一个变量（在本例中 game_x，作为分组依据，但这个变量只有两个值（1 和 0）。
我尝试了以下方法：
match_id <- c("match_1", "match_1","match_1","match_2","match_2","match_2","match_3","match_3","match_3", "match_4","match_4","match_4")
player_id <- c("player_1", "player_2", "player_3", "player_1", "player_2", "player_3", "player_1", "player_2", "player_3","player_1", "player_2", "player_3")
venue <- c("venue A", "venue A", "venue A", "venue B", "venue B", "venue B", "venue C", "venue C", "venue C","venue C", "venue C", "venue C")
opponent <- c("opponent A", "opponent A", "opponent A", "opponent B", "opponent B", "opponent B", "opponent C", "opponent C", "opponent C","opponent C", "opponent C", "opponent C")
game_x <- c(1,1,0,1,0,1,1,0,0,1,0,0)
points <- c(5,10,15,1,2,3,5,7,9,11,2,6)
data <- data.frame(match_id, player_id, venue, opponent, game_x, points)
library(data.table)
setDT(data)
f <- function(x,n=3) lapply(n:1,function(i) x[i:(i+length(x)-1)])
data[
order(player_id, game_x, -match_id),
c("last3", "last2", "last1"):=f(points,3),
by=.(player_id, game_x)][]
但它会产生这个：
match_id player_id venue opponent game_x points last3 last2 last1
1: match_1 player_1 venue A opponent A 1 5 NA NA 5
2: match_1 player_2 venue A opponent A 1 10 NA NA 10
3: match_1 player_3 venue A opponent A 0 15 NA NA 15
4: match_2 player_1 venue B opponent B 1 1 NA 5 1
5: match_2 player_2 venue B opponent B 0 2 NA NA 2
6: match_2 player_3 venue B opponent B 1 3 NA NA 3
7: match_3 player_1 venue C opponent C 1 5 5 1 5
8: match_3 player_2 venue C opponent C 0 7 NA 2 7
9: match_3 player_3 venue C opponent C 0 9 NA 15 9
10: match_4 player_1 venue C opponent C 1 11 1 5 11
11: match_4 player_2 venue C opponent C 0 2 2 7 2
12: match_4 player_3 venue C opponent C 0 6 15 9 6
什么时候应该是这样的：
match_id player_id venue opponent game_x points last3 last2 last1
1: match_4 player_1 venue C opponent C 1 11 1 5 11
2: match_3 player_1 venue C opponent C 1 5 5 1 5
3: match_2 player_1 venue B opponent B 1 1 NA 5 1
4: match_1 player_1 venue A opponent A 1 5 NA NA 5
5: match_4 player_2 venue C opponent C 0 2 NA NA NA
6: match_3 player_2 venue C opponent C 0 7 10 NA NA
7: match_2 player_2 venue B opponent B 0 2 NA 10 NA
8: match_1 player_2 venue A opponent A 1 10 NA NA 10
9: match_4 player_3 venue C opponent C 0 6 NA NA 3
10: match_3 player_3 venue C opponent C 0 9 NA 3 NA
11: match_2 player_3 venue B opponent B 1 3 NA NA 3
12: match_1 player_3 venue A opponent A 0 15 NA NA NA
我做错了什么？
更新二：
我试图跳过 game_x = 0 处的 NA（并将 game_x = 0 处的所有值替换为 NA）。
现在似乎正在发生的事情是，即使 game_x = 0，NA 也会出现在 last3、last2 等的列中。以最后 5 个为例：
match_id <- c(1,2,3,4,5)
player_id <- c("player_1", "player_1", "player_1", "player_1", "player_1")
venue <- c("venue A", "venue A", "venue B", "venue B", "venue B")
opponent <- c("opponent A", "opponent B", "opponent A", "opponent C", "opponent C")
game_x <- c(1,1,0,1,0)
points <- c(5,10,15,1,2)
data <- data.frame(match_id, player_id, venue, opponent, game_x, points)
library(data.table)
setDT(data)
f <- function(x,n=3,m=rep(TRUE,length(x))) {
x[!m] <- NA
lapply(n:1,function(i) x[i:(i+length(x)-1)])
}
data[
order(-match_id),
c("last5", "last4", "last3", "last2", "last1"):=f(points,5,game_x==1),
by=.(player_id)][order(player_id,-match_id)][]
data[order(-match_id)]
输出：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
1: 5 player_1 venue B opponent C 0 2 5 10 NA 1 NA
2: 4 player_1 venue B opponent C 1 1 NA 5 10 NA 1
3: 3 player_1 venue B opponent A 0 15 NA NA 5 10 NA
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5
应该看起来像：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
1: 5 player_1 venue B opponent C 0 2 NA NA NA NA NA
2: 4 player_1 venue B opponent C 1 1 NA NA 5 10 1
3: 3 player_1 venue B opponent A 0 15 NA NA NA NA NA
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5
您也许可以这样做：
- 设置数据为data.table
setDT(data)
- 创建一个 returns 向量列表的小函数，显示连续的最后
n 个点，给定一个向量作为输入
f <- function(x,n=3) lapply(n:1,\(i) x[i:(i+length(x)-1)])
- 按感兴趣的分组变量应用该函数，记住先排序。例如，要获得先验点数，只需通过玩家 ID，您可以使用
f()，如下所示：
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3),
by=player_id][]
- 如果您还想按场地和对手分组，请执行以下操作：
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3),
by=.(player_id, venue,opponent)][]
输出（player_id）：
match_id player_id venue opponent points last3 last2 last1
<char> <char> <char> <char> <num> <num> <num> <num>
1: match_1 player_1 venue A opponent A 5 NA NA 5
2: match_1 player_2 venue A opponent A 10 NA NA 10
3: match_1 player_3 venue A opponent A 15 NA NA 15
4: match_2 player_1 venue B opponent B 1 NA 5 1
5: match_2 player_2 venue B opponent B 2 NA 10 2
6: match_2 player_3 venue B opponent B 3 NA 15 3
7: match_3 player_1 venue C opponent C 5 5 1 5
8: match_3 player_2 venue C opponent C 7 10 2 7
9: match_3 player_3 venue C opponent C 9 15 3 9
10: match_4 player_1 venue C opponent C 11 1 5 11
11: match_4 player_2 venue C opponent C 2 2 7 2
12: match_4 player_3 venue C opponent C 6 3 9 6
如果你想要组合列，你可以这样做，假设你将上面的结果分配给r1
r1[, combined:=paste(last3,last2,last1,sep = ","), by=1:nrow(r1)][]
输出：
match_id player_id venue opponent points last3 last2 last1 combined
<char> <char> <char> <char> <num> <num> <num> <num> <char>
1: match_1 player_1 venue A opponent A 5 NA NA 5 NA,NA,5
2: match_1 player_2 venue A opponent A 10 NA NA 10 NA,NA,10
3: match_1 player_3 venue A opponent A 15 NA NA 15 NA,NA,15
4: match_2 player_1 venue B opponent B 1 NA 5 1 NA,5,1
5: match_2 player_2 venue B opponent B 2 NA 10 2 NA,10,2
6: match_2 player_3 venue B opponent B 3 NA 15 3 NA,15,3
7: match_3 player_1 venue C opponent C 5 5 1 5 5,1,5
8: match_3 player_2 venue C opponent C 7 10 2 7 10,2,7
9: match_3 player_3 venue C opponent C 9 15 3 9 15,3,9
10: match_4 player_1 venue C opponent C 11 1 5 11 1,5,11
11: match_4 player_2 venue C opponent C 2 2 7 2 2,7,2
12: match_4 player_3 venue C opponent C 6 3 9 6 3,9,6
这是所需的最少代码集：
library(data.table)
setDT(data)
f <- function(x,n=3) lapply(n:1,\(i) x[i:(i+length(x)-1)])
data[order(-match_id),c("last3", "last2", "last1"):=f(points,3),by=player_id]
data[, combined:=paste(last3,last2,last1,sep = ","), by=1:nrow(data)]
更新-
OP 现在想要在特定条件下排除某些行（跳过那些行）。如果可以将掩码传递给 f()，表示要包含哪些行，那么，我们可以这样调整 f()：
f <- function(x,n=3,m=rep(TRUE,length(x))) {
x[!m] <- NA
lapply(n:1,function(i) x[i:(i+length(x)-1)])
}
此示例使用上述 f() 的调整版本来跳过 game_x==0
的行
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3,game_x==1),
by=.(player_id)][order(player_id,-match_id)][]
另一个更新！，
现在 OP 想要完全排除 game_x=0 行。
rbind(
data[game_x==0],
data[game_x==1][
order(-match_id),
c("last5", "last4", "last3", "last2", "last1"):=f(points,5),
by=.(player_id)][order(player_id,-match_id)],
fill=TRUE
)
输出：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
<num> <char> <char> <char> <num> <num> <num> <num> <num> <num> <num>
1: 3 player_1 venue B opponent A 0 15 NA NA NA NA NA
2: 5 player_1 venue B opponent C 0 2 NA NA NA NA NA
3: 4 player_1 venue B opponent C 1 1 NA NA 5 10 1
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5
我以下面的数据框为例：
match_id <- c("match_1", "match_1","match_1","match_2","match_2","match_2","match_3","match_3","match_3", "match_4","match_4","match_4")
player_id <- c("player_1", "player_2", "player_3", "player_1", "player_2", "player_3", "player_1", "player_2", "player_3","player_1", "player_2", "player_3")
venue <- c("venue A", "venue A", "venue A", "venue B", "venue B", "venue B", "venue C", "venue C", "venue C","venue C", "venue C", "venue C")
opponent <- c("opponent A", "opponent A", "opponent A", "opponent B", "opponent B", "opponent B", "opponent C", "opponent C", "opponent C","opponent C", "opponent C", "opponent C")
points <- c(5,10,15,1,2,3,5,7,9,11,2,6)
data <- data.frame(match_id, player_id, venue, opponent, points)
我希望创建新的列，以根据分组显示点数列中的最后 n 个值。
第一组我的尝试，以3局为例：
#group by player, give the last n games
library(dplyr)
data <- data %>%
arrange(player_id,desc(match_id)) %>%
mutate(last3_games = lag(points, n=3)) %>%
mutate(last2_games = lag(points, n=2)) %>%
mutate(last1_games = lag(points, n=1))
给出：
head(data)
match_id player_id venue opponent points last3_games last2_games last1_games
1 match_4 player_1 venue C opponent C 11 NA NA NA
2 match_3 player_1 venue C opponent C 6 NA NA 11
3 match_2 player_1 venue B opponent B 1 NA 11 5
4 match_1 player_1 venue A opponent A 5 11 5 1
5 match_4 player_2 venue C opponent C 2 5 1 5
6 match_3 player_2 venue C opponent C 7 1 5 2
7 match_2 player_2 venue B opponent B 3 5 2 7
8 match_1 player_2 venue A opponent A 10 2 7 2
但我希望显示的值是每个玩家的最后 3 个点值，如下所示：
match_id player_id venue opponent points last3_games last2_games last1_games
1 match_4 player_1 venue C opponent C 11 1 6 11
2 match_3 player_1 venue C opponent C 6 5 1 6
3 match_2 player_1 venue B opponent B 1 NA 5 1
4 match_1 player_1 venue A opponent A 5 NA NA 5
5 match_4 player_2 venue C opponent C 2 3 7 2
6 match_3 player_2 venue C opponent C 7 10 3 7
7 match_2 player_2 venue B opponent B 3 NA 10 3
8 match_1 player_2 venue A opponent A 10 NA NA 10
然后我也想通过对手和场地做同样的事情：
#by opponent
match_id player_id venue opponent points last3_opponent last2_opponent last1_opponent
1 match_4 player_1 venue C opponent C 11 NA 5 11
2 match_3 player_1 venue C opponent C 5 NA NA 5
3 match_2 player_1 venue B opponent B 1 NA NA 1
4 match_1 player_1 venue A opponent A 5 NA NA 5
5 match_4 player_2 venue C opponent C 2 NA 7 2
6 match_3 player_2 venue C opponent C 7 NA NA 7
7 match_2 player_2 venue B opponent B 2 NA NA 2
8 match_1 player_2 venue A opponent A 10 NA NA 10
这里有很多NA's，但这是因为我只展示了一个小例子。
我的尝试是再次将它们分组并使用滞后进行变异，但它没有提供所需的输出：
data <- data %>%
arrange(player_id,desc(match_id), opponent) %>%
mutate(last3_opponent = lag(points, n=3)) %>%
mutate(last2_opponent = lag(points, n=2)) %>%
mutate(last1_opponent = lag(points, n=1))
按对手和场地分组的输出应如下所示：
#by opponent and venue
match_id player_id venue opponent points last3_opp_ven last2_opp_ven last1_opp_ven
1 match_1 player_1 venue A opponent A 5 NA NA 5
2 match_2 player_1 venue B opponent B 1 NA NA 1
3 match_4 player_1 venue C opponent C 11 NA 5 11
4 match_3 player_1 venue C opponent C 5 NA NA 5
5 match_1 player_2 venue A opponent A 10 NA NA 10
6 match_2 player_2 venue B opponent B 2 NA NA 2
7 match_4 player_2 venue C opponent C 2 NA 7 2
8 match_3 player_2 venue C opponent C 7 NA NA 7
理想情况下，我正在寻找最近 10 场比赛（最后 10 场、最后 9 场、最后 8 场等），并且由于有很多对手和很多 vanues，因此生成的数据框将有很多列。
有没有更简单的方法？
我还想在每个组的末尾添加一列，其中包含所有值的组合，对于每个组，就像这样（使用游戏的第一个示例）：
match_id player_id venue opponent points last3_combined
1 match_4 player_1 venue C opponent C 11 1,6,11
2 match_3 player_1 venue C opponent C 6 5,1,6
3 match_2 player_1 venue B opponent B 1 NA,5,1
4 match_1 player_1 venue A opponent A 5 NA,NA,5
5 match_4 player_2 venue C opponent C 2 3,7,2
6 match_3 player_2 venue C opponent C 7 10,3,7
7 match_2 player_2 venue B opponent B 3 NA,10,3
8 match_1 player_2 venue A opponent A 10 NA,NA,10
更新：
假设我想使用另一个变量（在本例中 game_x，作为分组依据，但这个变量只有两个值（1 和 0）。
我尝试了以下方法：
match_id <- c("match_1", "match_1","match_1","match_2","match_2","match_2","match_3","match_3","match_3", "match_4","match_4","match_4")
player_id <- c("player_1", "player_2", "player_3", "player_1", "player_2", "player_3", "player_1", "player_2", "player_3","player_1", "player_2", "player_3")
venue <- c("venue A", "venue A", "venue A", "venue B", "venue B", "venue B", "venue C", "venue C", "venue C","venue C", "venue C", "venue C")
opponent <- c("opponent A", "opponent A", "opponent A", "opponent B", "opponent B", "opponent B", "opponent C", "opponent C", "opponent C","opponent C", "opponent C", "opponent C")
game_x <- c(1,1,0,1,0,1,1,0,0,1,0,0)
points <- c(5,10,15,1,2,3,5,7,9,11,2,6)
data <- data.frame(match_id, player_id, venue, opponent, game_x, points)
library(data.table)
setDT(data)
f <- function(x,n=3) lapply(n:1,function(i) x[i:(i+length(x)-1)])
data[
order(player_id, game_x, -match_id),
c("last3", "last2", "last1"):=f(points,3),
by=.(player_id, game_x)][]
但它会产生这个：
match_id player_id venue opponent game_x points last3 last2 last1
1: match_1 player_1 venue A opponent A 1 5 NA NA 5
2: match_1 player_2 venue A opponent A 1 10 NA NA 10
3: match_1 player_3 venue A opponent A 0 15 NA NA 15
4: match_2 player_1 venue B opponent B 1 1 NA 5 1
5: match_2 player_2 venue B opponent B 0 2 NA NA 2
6: match_2 player_3 venue B opponent B 1 3 NA NA 3
7: match_3 player_1 venue C opponent C 1 5 5 1 5
8: match_3 player_2 venue C opponent C 0 7 NA 2 7
9: match_3 player_3 venue C opponent C 0 9 NA 15 9
10: match_4 player_1 venue C opponent C 1 11 1 5 11
11: match_4 player_2 venue C opponent C 0 2 2 7 2
12: match_4 player_3 venue C opponent C 0 6 15 9 6
什么时候应该是这样的：
match_id player_id venue opponent game_x points last3 last2 last1
1: match_4 player_1 venue C opponent C 1 11 1 5 11
2: match_3 player_1 venue C opponent C 1 5 5 1 5
3: match_2 player_1 venue B opponent B 1 1 NA 5 1
4: match_1 player_1 venue A opponent A 1 5 NA NA 5
5: match_4 player_2 venue C opponent C 0 2 NA NA NA
6: match_3 player_2 venue C opponent C 0 7 10 NA NA
7: match_2 player_2 venue B opponent B 0 2 NA 10 NA
8: match_1 player_2 venue A opponent A 1 10 NA NA 10
9: match_4 player_3 venue C opponent C 0 6 NA NA 3
10: match_3 player_3 venue C opponent C 0 9 NA 3 NA
11: match_2 player_3 venue B opponent B 1 3 NA NA 3
12: match_1 player_3 venue A opponent A 0 15 NA NA NA
我做错了什么？
更新二：
我试图跳过 game_x = 0 处的 NA（并将 game_x = 0 处的所有值替换为 NA）。
现在似乎正在发生的事情是，即使 game_x = 0，NA 也会出现在 last3、last2 等的列中。以最后 5 个为例：
match_id <- c(1,2,3,4,5)
player_id <- c("player_1", "player_1", "player_1", "player_1", "player_1")
venue <- c("venue A", "venue A", "venue B", "venue B", "venue B")
opponent <- c("opponent A", "opponent B", "opponent A", "opponent C", "opponent C")
game_x <- c(1,1,0,1,0)
points <- c(5,10,15,1,2)
data <- data.frame(match_id, player_id, venue, opponent, game_x, points)
library(data.table)
setDT(data)
f <- function(x,n=3,m=rep(TRUE,length(x))) {
x[!m] <- NA
lapply(n:1,function(i) x[i:(i+length(x)-1)])
}
data[
order(-match_id),
c("last5", "last4", "last3", "last2", "last1"):=f(points,5,game_x==1),
by=.(player_id)][order(player_id,-match_id)][]
data[order(-match_id)]
输出：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
1: 5 player_1 venue B opponent C 0 2 5 10 NA 1 NA
2: 4 player_1 venue B opponent C 1 1 NA 5 10 NA 1
3: 3 player_1 venue B opponent A 0 15 NA NA 5 10 NA
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5
应该看起来像：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
1: 5 player_1 venue B opponent C 0 2 NA NA NA NA NA
2: 4 player_1 venue B opponent C 1 1 NA NA 5 10 1
3: 3 player_1 venue B opponent A 0 15 NA NA NA NA NA
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5
您也许可以这样做：
- 设置数据为data.table
setDT(data)
- 创建一个 returns 向量列表的小函数，显示连续的最后
n个点，给定一个向量作为输入
f <- function(x,n=3) lapply(n:1,\(i) x[i:(i+length(x)-1)])
- 按感兴趣的分组变量应用该函数，记住先排序。例如，要获得先验点数，只需通过玩家 ID，您可以使用
f()，如下所示：
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3),
by=player_id][]
- 如果您还想按场地和对手分组，请执行以下操作：
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3),
by=.(player_id, venue,opponent)][]
输出（player_id）：
match_id player_id venue opponent points last3 last2 last1
<char> <char> <char> <char> <num> <num> <num> <num>
1: match_1 player_1 venue A opponent A 5 NA NA 5
2: match_1 player_2 venue A opponent A 10 NA NA 10
3: match_1 player_3 venue A opponent A 15 NA NA 15
4: match_2 player_1 venue B opponent B 1 NA 5 1
5: match_2 player_2 venue B opponent B 2 NA 10 2
6: match_2 player_3 venue B opponent B 3 NA 15 3
7: match_3 player_1 venue C opponent C 5 5 1 5
8: match_3 player_2 venue C opponent C 7 10 2 7
9: match_3 player_3 venue C opponent C 9 15 3 9
10: match_4 player_1 venue C opponent C 11 1 5 11
11: match_4 player_2 venue C opponent C 2 2 7 2
12: match_4 player_3 venue C opponent C 6 3 9 6
如果你想要组合列，你可以这样做，假设你将上面的结果分配给r1
r1[, combined:=paste(last3,last2,last1,sep = ","), by=1:nrow(r1)][]
输出：
match_id player_id venue opponent points last3 last2 last1 combined
<char> <char> <char> <char> <num> <num> <num> <num> <char>
1: match_1 player_1 venue A opponent A 5 NA NA 5 NA,NA,5
2: match_1 player_2 venue A opponent A 10 NA NA 10 NA,NA,10
3: match_1 player_3 venue A opponent A 15 NA NA 15 NA,NA,15
4: match_2 player_1 venue B opponent B 1 NA 5 1 NA,5,1
5: match_2 player_2 venue B opponent B 2 NA 10 2 NA,10,2
6: match_2 player_3 venue B opponent B 3 NA 15 3 NA,15,3
7: match_3 player_1 venue C opponent C 5 5 1 5 5,1,5
8: match_3 player_2 venue C opponent C 7 10 2 7 10,2,7
9: match_3 player_3 venue C opponent C 9 15 3 9 15,3,9
10: match_4 player_1 venue C opponent C 11 1 5 11 1,5,11
11: match_4 player_2 venue C opponent C 2 2 7 2 2,7,2
12: match_4 player_3 venue C opponent C 6 3 9 6 3,9,6
这是所需的最少代码集：
library(data.table)
setDT(data)
f <- function(x,n=3) lapply(n:1,\(i) x[i:(i+length(x)-1)])
data[order(-match_id),c("last3", "last2", "last1"):=f(points,3),by=player_id]
data[, combined:=paste(last3,last2,last1,sep = ","), by=1:nrow(data)]
更新-
OP 现在想要在特定条件下排除某些行（跳过那些行）。如果可以将掩码传递给 f()，表示要包含哪些行，那么，我们可以这样调整 f()：
f <- function(x,n=3,m=rep(TRUE,length(x))) {
x[!m] <- NA
lapply(n:1,function(i) x[i:(i+length(x)-1)])
}
此示例使用上述 f() 的调整版本来跳过 game_x==0
data[
order(-match_id),
c("last3", "last2", "last1"):=f(points,3,game_x==1),
by=.(player_id)][order(player_id,-match_id)][]
另一个更新！，
现在 OP 想要完全排除 game_x=0 行。
rbind(
data[game_x==0],
data[game_x==1][
order(-match_id),
c("last5", "last4", "last3", "last2", "last1"):=f(points,5),
by=.(player_id)][order(player_id,-match_id)],
fill=TRUE
)
输出：
match_id player_id venue opponent game_x points last5 last4 last3 last2 last1
<num> <char> <char> <char> <num> <num> <num> <num> <num> <num> <num>
1: 3 player_1 venue B opponent A 0 15 NA NA NA NA NA
2: 5 player_1 venue B opponent C 0 2 NA NA NA NA NA
3: 4 player_1 venue B opponent C 1 1 NA NA 5 10 1
4: 2 player_1 venue A opponent B 1 10 NA NA NA 5 10
5: 1 player_1 venue A opponent A 1 5 NA NA NA NA 5