R 分组为循环 seq_along 长度 - 1？

Question

我正在尝试创建一个指标，用于指示项目被接受所需的“尝试”次数。我认为 for 循环是要走的路，但我在 R 中没有大量的循环经验，而且逻辑有点复杂。任何帮助/建议/反馈将不胜感激！

在玩具示例中，“accept”为“C”，向前迭代“try”的切换是提交 (A) 是否被重置 (B) 或提交 (A) 是否被接受 (C)。

在一个组内，如果事件顺序是 A > B 或 A > C，则将“try”向前迭代 1。否则，“try”计数应保持不变。显然，“真实”的例子比这个玩具例子要复杂得多。

目前，我只是想使尝试计数正确，而不用担心分组。

我不确定如何将 seq_along 限制为基本上停止在 [group_by %>% length(group) - 1]。有更好的选择吗？

df = data.frame(group = c(1,1,1,1,1,2,2,2,2), 
                 event = c("A","B","A","A","C","A","B","A","C"))

df$try <- 0
for (i in seq_along(df$event)){
    if (df$event[[i]] == "A" &  
          df$event[[i+1]] %in% c("B", "C"))
      {
        df$try[[i]] <- df$try + 1
    } else {
        df$try[[i]] <- df$try
    }
}

# this essentially shows the correct answer (win = try + 1, loss = try), 
# but has "df$event[[i + 1]] : subscript out of bounds", 
# and I need to save the outcome so I can access later

df$try <- 0
for (i in seq_along(df$event)){
    if (df$event[[i]] == "A" &  
          df$event[[i+1]] %in% c("B", "C"))
      {
        print("Win")
    } else {
        print("Loss")
    }
}

我对玩具示例的预期（最终）答案是：try = c(1,1,1,2,2,1,1,2,2);每组 1 和 2 需要 2 次“尝试”才能被接受

Answer 1

您可以使用lead 获取dplyr 中的下一个值。试试这个 -

library(dplyr)

df %>%
  group_by(group) %>%
  mutate(result = cumsum(event == 'A' & lead(event) %in% c('B', 'C'))) %>%
  ungroup

#  group event   try result
#  <dbl> <chr> <dbl>  <int>
#1     1 A         1      1
#2     1 B         1      1
#3     1 A         1      1
#4     1 A         2      2
#5     1 C         2      2
#6     2 A         1      1
#7     2 B         1      1
#8     2 A         2      2
#9     2 C         2      2

在输出中保留try 列以供比较。

Answer 2

多加一个if可以解决“下标越界”的问题。

if(i+1 > nrow(df){
print('do nothing')
} else if (
#followed by your original code
)

我假设如果最后一行，值将只是 0。所以另一个 if 应该这样做。

library(tidyverse)
df <- data.frame(group = c(1,1,1,1,1,2,2,2,2), 
                event = c("A","B","A","A","C","A","B","A","C"))


temp <- data.frame(NULL)
for(i in 1:nrow(df)){
  if(i+1 > nrow(df)){
    print('This is the last row')
    temp <- rbind(temp, 0)
  } else if(df$event[[i]] == 'A' &
     df$event[[i+1]] %in% c('B', 'C'))
  {
    temp <- rbind(temp, 1)
  } else {
    temp <- rbind(temp, 0)
  }
}

df2 <- cbind(df, temp) %>%
  mutate(
    cumulative_sum = cumsum(X1)
  )

Answer 3

这似乎暂时有效：

如果 i + 1 超出长度，则添加“中断”

df$try <- 0
for (i in seq_along(df$event)){
    if (i+1 == length(df$event)){
      break
      } else if (df$event[[i]] == "A" &  
          df$event[[i+1]] %in% c("B", "C"))
      {
        print("Win")
    } else (
        print("Loss")
    )
}

# updated toy df to show N tries differs:

df = data.frame(group = c(1,1,1,1,1,1,1,2,2,2,2), 
                 event = c("A","B","A","A","B","A","C","A","B","A","C"))

df$try <- 0
for (i in seq_along(df$event)){
    if (i == length(df$event)){ # use i otherwise it doesn't catch the last switch
      break
      } else if (df$event[[i]] == "A" &  
          df$event[[i+1]] %in% c("B", "C"))
      {
        df$try[[i]] <- + 1
    } else (
        df$try[[i]]
    )
}

df %>% 
  group_by(group) %>% 
  mutate(N_tries = max(cumsum(try)))