R中唯一game_id的不同变量的Cumsum答案

【问题标题】：Cumsum for different variables for unique game_id in RR中唯一game_id的不同变量的Cumsum
【发布时间】：2021-04-17 11:34:03
【问题描述】：

我有以下数据框，其中包含游戏 ID、玩家、动作类型（例如传球或运球）以及动作是成功还是失败。

df1 <- data.frame(
  game_id = c("1", "1", "1", "1", "1", "1", "1", "1", "1", "1", "2", "2", "2", "2", "2", 
              "2", "2", "2", "2", "2"),
  player  = c("X", "X", "X", "Y", "Y", "Z", "Z", "X", "Y", "Z", "Y", "Y", "Y", "X", "X", "Z",
              "Z", "X", "Z", "X"),
  type    = c("pass", "pass", "pass", "pass", "pass", "dribble", "dribble", 'tackle', "pass",
              "pass", "dribble", "pass", "dribble", "pass", "pass", "pass", "pass", "dribble", 
              "pass", "pass"),
  result  = c("success", "success", "fail", "success", "success", "success", "fail", "success",
              "fail", "success", "success", "fail", "fail", "success", "success", "fail", "fail",
              "success", "success", "success")
)

df1

#   game_id player    type  result
#1        1      X    pass success
#2        1      X    pass success
#3        1      X    pass    fail
#4        1      Y    pass success
#5        1      Y    pass success
#6        1      Z dribble success
#7        1      Z dribble    fail
#8        1      X  tackle success
#9        1      Y    pass    fail
#10       1      Z    pass success
#11       2      Y dribble success
#12       2      Y    pass    fail
#13       2      Y dribble    fail
#14       2      X    pass success
#15       2      X    pass success
#16       2      Z    pass    fail
#17       2      Z    pass    fail
#18       2      X dribble success
#19       2      Z    pass success
#20       2      X    pass success

我想做的是创建一些新列：

第一列包含玩家在独特比赛中的每一次传球（无论传球失败还是成功）
第二列包含玩家在一场独特的比赛中给出的每个 SUCCESS 通行证
第三列显示每个球员在比赛中的成功率（因此球员的成功传球次数/球员传球总数）

最终的数据框应如下所示：

df2 <- data.frame(
  game_id = c(1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2),
  player  = c("X", "X", "X", "Y", "Y", "Z", "Z", "X", "Y", "Z", "Y", "Y", "Y", "X", "X", "Z",
              "Z", "X", "Z", "X"),
  type    = c("pass", "pass", "pass", "pass", "pass", "dribble", "dribble", 'tackle', "pass",
              "pass","dribble", "pass", "dribble", "pass", "pass", "pass", "pass", "dribble", 
              "pass", "pass"),
  result  = c("success", "success", "fail", "success", "success", "success", "fail", "success",
              "fail", "success", "success", "fail", "fail", "success", "success", "fail", "fail",
              "success", "success", "success"),
  pass_per_player = c(1,2,3,1,2,0,0,3,3,1,0,1,1,1,2,1,2,2,3,3), 
  success_pass_player = c(1,2,2,1,2,0,0,2,2,1,0,0,0,1,2,0,0,2,1,3),
  #success_rate_player = c(1,1,0.66,1,1,0,0,0.66,0.66,1,0,0,0,1,1,0,0,1,0.33,1)
)

df2

#   game_id player    type  result pass_per_player success_pass_player success_rate_player
#1        1      X    pass success               1                   1                1.00
#2        1      X    pass success               2                   2                1.00
#3        1      X    pass    fail               3                   2                0.66
#4        1      Y    pass success               1                   1                1.00
#5        1      Y    pass success               2                   2                1.00
#6        1      Z dribble success               0                   0                0.00
#7        1      Z dribble    fail               0                   0                0.00
#8        1      X  tackle success               3                   2                0.66
#9        1      Y    pass    fail               3                   2                0.66
#10       1      Z    pass success               1                   1                1.00
#11       2      Y dribble success               0                   0                0.00
#12       2      Y    pass    fail               1                   0                0.00
#13       2      Y dribble    fail               1                   0                0.00
#14       2      X    pass success               1                   1                1.00
#15       2      X    pass success               2                   2                1.00
#16       2      Z    pass    fail               1                   0                0.00
#17       2      Z    pass    fail               2                   0                0.00
#18       2      X dribble success               2                   2                1.00
#19       2      Z    pass success               3                   1                0.33
#20       2      X    pass success               3                   3                1.00

【问题讨论】：

标签： r

【解决方案1】：

第一个开始是这样的：

df1 %>%
  group_by(game_id, player) %>%
  mutate(
    pass_per_player = cumsum(type=="pass"),
    success_pass_player = cumsum(result=="success" & type=="pass"),
    success_rate_player = success_pass_player / pass_per_player)
# A tibble: 20 x 7
# Groups:   game_id, player [6]
   game_id player type    result  pass_per_player success_pass_player success_rate_player
   <chr>   <chr>  <chr>   <chr>             <int>               <int>               <dbl>
 1 1       X      pass    success               1                   1               1    
 2 1       X      pass    success               2                   2               1    
 3 1       X      pass    fail                  3                   2               0.667
 4 1       Y      pass    success               1                   1               1    
 5 1       Y      pass    success               2                   2               1    
 6 1       Z      dribble success               0                   0             NaN    
 7 1       Z      dribble fail                  0                   0             NaN    
 8 1       X      tackle  success               3                   2               0.667
 9 1       Y      pass    fail                  3                   2               0.667
10 1       Z      pass    success               1                   1               1    
11 2       Y      dribble success               0                   0             NaN    
12 2       Y      pass    fail                  1                   0               0    
13 2       Y      dribble fail                  1                   0               0    
14 2       X      pass    success               1                   1               1    
15 2       X      pass    success               2                   2               1    
16 2       Z      pass    fail                  1                   0               0    
17 2       Z      pass    fail                  2                   0               0    
18 2       X      dribble success               2                   2               1    
19 2       Z      pass    success               3                   1               0.333
20 2       X      pass    success               3                   3               1

如果要将NaNs 转换为0：

df1$success_rate_player[is.nan(df1$success_rate_player)] <- 0

【讨论】：

但是，我相信第一个 cumsum 并没有真正起作用，因为它会增加每次传球，而不是每个玩家的传球
你想要的结果是一样的
不幸的是，上面的代码给了我不同的输出。我现在尝试了以下方法：
df1 <- ddply(df1, .(game_id, player), transform, pass_per_player = cumsum(type == "pass"))