如何在没有循环的情况下将频率表汇总为更小的间隔？

Question

我想通过按变量和特定间隔过滤而不使用循环并手动创建大部分数据框，将频率表汇总为更紧凑的频率表。

我在做什么：

head(ride_duration)
  member_casual   duration frequency
1        casual 0.00000000       235
2        member 0.00000000       228
3        casual 0.01666667       567
4        member 0.01666667       813
5        casual 0.03333333      1527
6        member 0.03333333      3570
...

tail(ride_duration)
      member_casual duration frequency
45193        casual 35820.63         1
45194        member 35820.63         0
45195        casual 36257.80         1
45196        member 36257.80         0
45197        casual 40705.02         1
45198        member 40705.02         0

期望的结果：

   member_casual        interval frequency
1         member      0-1 minute   sum of duration <=1
2         member     1-5 minutes   sum of duration >1<=5
3         member    5-10 minutes   sum of duration >5<=10
4         member   10-15 minutes         .
5         member   15-30 minutes         .
6         member   30-60 minutes         .
7         member  60-120 minutes         .
8         member  120-180 minute         .
9         member 180-240 minutes         .
10        member    240+ minutes         .
11        casual      0-1 minute   sum of duration <=1
12        casual     1-5 minutes   sum of duration >1<=5
13        casual    5-10 minutes   sum of duration >5<=10
14        casual   10-15 minutes         .
15        casual   15-30 minutes         .
16        casual   30-60 minutes         .
17        casual  60-120 minutes         .
18        casual  120-180 minute         .
19        casual 180-240 minutes         .
20        casual    240+ minutes         .

我能够使用循环解决这个问题，这花了我很长时间才弄清楚，但我觉得应该有一个函数可以使这个过程变得更加容易。这是我最终所做的工作：

# Create table with set minute intervals with frequencies
ride_duration_pyramid <- data.frame("member_casual"=c("casual","casual","casual","casual","casual","casual","casual","casual","casual","casual",
                                                      "member","member","member","member","member","member","member","member","member","member"),
                                    "interval"=c("0-1 minute", "1-5 minutes", "5-10 minutes", "10-15 minutes", "15-30 minutes",
                                                 "30-60 minutes","60-120 minutes","120-180 minute", "180-240 minutes", "240+ minutes",
                                                 "0-1 minute", "1-5 minutes", "5-10 minutes", "10-15 minutes", "15-30 minutes",
                                                 "30-60 minutes","60-120 minutes","120-180 minute", "180-240 minutes", "240+ minutes"),
                                    "frequency"=replicate(20,0))
current_interval <- 1 # Starting interval
interval_high <- c(1,5,10,15,30,60,120,180,240,40706) # Interval maxs
for (i in 1:nrow(ride_duration)) { # Scan all rows
  row <- ride_duration[i,]
  # Walk through intervals to find the row this data point falls into.
  while (row[2] >= interval_high[current_interval]) {
    current_interval <- current_interval + 1;
  }
  if (row[1] == "casual") {
    ride_duration_pyramid[current_interval,3] <- ride_duration_pyramid[current_interval,3] + row[3];
  } else {
    ride_duration_pyramid[current_interval+10,3] <- ride_duration_pyramid[current_interval+10,3] + row[3];
  }
}

Answer 1

这是cut和aggregate的一种方式。

• 定义断点向量；
• 定义一个分箱数据标签向量；
• cut断点持续时间的数字向量并设置因子水平；
• 最后，aggregate数据。

请注意，中断向量与问题中发布的不完全相同。它不是最大值向量，它从可能的最低值开始，到可能的最大值结束。

ride_duration <- "
member_casual   duration frequency
1        casual 0.00000000       235
2        member 0.00000000       228
3        casual 0.01666667       567
4        member 0.01666667       813
5        casual 0.03333333      1527
6        member 0.03333333      3570
45193        casual 35820.63         1
45194        member 35820.63         0
45195        casual 36257.80         1
45196        member 36257.80         0
45197        casual 40705.02         1
45198        member 40705.02         0
"
ride_duration <- read.table(textConnection(ride_duration), header = TRUE)

#
interval_high <- c(0,1,5,10,15,30,60,120,180,240,Inf)
interval_labels <- c("0-1 minute", "1-5 minutes", "5-10 minutes", "10-15 minutes", 
                     "15-30 minutes", "30-60 minutes", "60-120 minutes", "120-180 minute", 
                     "180-240 minutes", "240+ minutes")
interval <- cut(ride_duration$duration, interval_high, labels = interval_labels, include.lowest = TRUE)

aggregate(frequency ~ interval + member_casual, ride_duration, sum, na.rm = TRUE)[c(2:1, 3)]
#>   member_casual     interval frequency
#> 1        casual   0-1 minute      2329
#> 2        casual 240+ minutes         3
#> 3        member   0-1 minute      4611
#> 4        member 240+ minutes         0

^{创建于 2022-12-05 reprex v2.0.2}