Question

我有两个数据集：一个带有航点，另一个带有航迹。

我想根据跟踪的时间将航迹数据集的“模式”变量添加到航点数据集中

在航点数据集中，我有变量“ tracked_at”

`

ID <- c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15, 16, 17, 18, 19, 20)
tracked_at <- ymd_hms(c("2017-10-16 06:00:02", "2017-10-16 06:00:07", "2017-10-16 06:01:15", "2017-10-16 06:02:09",
                     "2017-10-16 06:02:50", "2017-10-16 06:04:05", "2017-10-16 06:04:15", "2017-10-16 06:10:15",
                     "2017-10-16 06:14:15", "2017-10-16 06:16:15", "2017-10-16 06:18:30", "2017-10-16 06:18:45", 
                     "2017-10-16 06:19:15", "2017-10-16 06:19:40", "2017-10-16 06:19:55", "2017-10-17 08:08:02", 
                     "2017-10-17 08:10:02", "2017-10-17 08:16:02", "2017-10-17 08:17:02", "2017-10-18 15:00:00"))

lat <- c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)
long <- c(2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2)
id_user<- c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,3)
df1 <- data.frame(ID, tracked_at,lat, long, id_user )

print(df1)
   ID          tracked_at lat long id_user
1   1 2017-10-16 06:00:02   1    2       1
2   2 2017-10-16 06:00:07   1    2       1
3   3 2017-10-16 06:01:15   1    2       1
4   4 2017-10-16 06:02:09   1    2       1
5   5 2017-10-16 06:02:50   1    2       1
6   6 2017-10-16 06:04:05   1    2       1
7   7 2017-10-16 06:04:15   1    2       1
8   8 2017-10-16 06:10:15   1    2       1
9   9 2017-10-16 06:14:15   1    2       1
10 10 2017-10-16 06:16:15   1    2       1
11 11 2017-10-16 06:18:30   1    2       1
12 12 2017-10-16 06:18:45   1    2       1
13 13 2017-10-16 06:19:15   1    2       1
14 14 2017-10-16 06:19:40   1    2       1
15 15 2017-10-16 06:19:55   1    2       1
16 16 2017-10-17 08:08:02   1    2       2
17 17 2017-10-17 08:10:02   1    2       2
18 18 2017-10-17 08:16:02   1    2       2
19 19 2017-10-17 08:17:02   1    2       2
20 20 2017-10-18 15:00:00   1    2       3

在轨迹数据集中，我有变量“ started_at”和“ finished_a”

started_at <- ymd_hms(c("2017-10-16 06:00:05", "2017-10-16 06:04:15", "2017-10-16 06:18:31", "2017-10-17 08:10:02"))
finished_a <- ymd_hms(c("2017-10-16 06:02:10", "2017-10-16 06:18:30", "2017-10-16 06:19:45", "2017-10-17 08:16:02"))
id_user <- c(1, 1, 1, 2)
Mode <- c("Walk", "Train", "Walk", "Car")
df2 <- data.frame(started_at,finished_a, id_user, Mode )
print(df2)

           started_at          finished_a id_user  Mode
1 2017-10-16 06:00:05 2017-10-16 06:02:10       1  Walk
2 2017-10-16 06:04:15 2017-10-16 06:18:30       1 Train
3 2017-10-16 06:18:31 2017-10-16 06:19:45       1  Walk
4 2017-10-17 08:10:02 2017-10-17 08:16:02       2   Car

这3个变量是日期格式（ymd_hms），当个人不动时也会跟踪航路点，因此模式列应在大多数时间由NA填充。我想补充一点，如果NA在“巴士”或“火车”模式之前以及“步行”模式之后，那么它就是“等待时间”。

这将是理想的数据集：

ID <- c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15, 16, 17, 18, 19, 20)
tracked_at <- ymd_hms(c("2017-10-16 06:00:02", "2017-10-16 06:00:07", "2017-10-16 06:01:15", "2017-10-16 06:02:09",
                 "2017-10-16 06:02:50", "2017-10-16 06:04:05", "2017-10-16 06:04:15", "2017-10-16 06:10:15",
                 "2017-10-16 06:14:15", "2017-10-16 06:16:15", "2017-10-16 06:18:30", "2017-10-16 06:18:45", 
                 "2017-10-16 06:19:15", "2017-10-16 06:19:40", "2017-10-16 06:19:55", "2017-10-17 08:08:02", 
                 "2017-10-17 08:10:02", "2017-10-17 08:16:02", "2017-10-17 08:17:02", "2017-10-18 15:00:00"))

lat <- c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)
long <- c(2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2)
id_user<- c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,3)
NeWmode <- c("NA","Walk", "Walk", "Walk", "Waiting Time", "Waiting Time", "Train", "Train", "Train", "Train", "Train",
             "Walk","Walk", "Walk", "NA", "NA", "Car", "Car", "NA", "NA")
df3 <- data.frame(ID, tracked_at,lat, long, id_user, Newmode )
print(df3)

ID              tracked_at lat long id_user      Newmode
1   1 2017-10-16 06:00:02   1    2       1           NA
2   2 2017-10-16 06:00:07   1    2       1         Walk
3   3 2017-10-16 06:01:15   1    2       1         Walk
4   4 2017-10-16 06:02:09   1    2       1         Walk
5   5 2017-10-16 06:02:50   1    2       1 Waiting Time
6   6 2017-10-16 06:04:05   1    2       1 Waiting Time
7   7 2017-10-16 06:04:15   1    2       1        Train
8   8 2017-10-16 06:10:15   1    2       1        Train
9   9 2017-10-16 06:14:15   1    2       1        Train
10 10 2017-10-16 06:16:15   1    2       1        Train
11 11 2017-10-16 06:18:30   1    2       1        Train
12 12 2017-10-16 06:18:45   1    2       1         Walk
13 13 2017-10-16 06:19:15   1    2       1         Walk
14 14 2017-10-16 06:19:40   1    2       1         Walk
15 15 2017-10-16 06:19:55   1    2       1           NA
16 16 2017-10-17 08:08:02   1    2       2           NA
17 17 2017-10-17 08:10:02   1    2       2          Car
18 18 2017-10-17 08:16:02   1    2       2          Car
19 19 2017-10-17 08:17:02   1    2       2           NA
20 20 2017-10-18 15:00:00   1    2       3           NA

我最好的猜测是：

   id1 <-filter(df1, id_user==1) #filtering by users
id1Moda <- filter(df2, id_user==1)

id1Moda$mode.num[id1Moda$Mode=="Walk"] <-1
id1Moda$mode.num[id1Moda$Mode=="Train"] <-2
id1Moda$mode.num[id1Moda$Mode=="Car"] <-3

  id1$mode <- NA
for(i in 1:nrow(id1Moda)){
  for(k in 1:nrow(id1)){
    if((id1$tracked_at[k] >= id1Moda$started_at[i]) & (id1$tracked_at[k] <= id1Moda$finished_a[i])){
      id1$Newmode[k] <- id1Moda$mode.num[i]
    } else {
    }
  }
}

理想情况下，我没有按用户过滤，因为我有50位用户进行分析而且，循环非常慢，因为该ID的数据集超过280 000点。整个数据库总共获得了38000000个数据点。

备注：

记录航路点时不一定会开始跟踪，这就是我使用间隔的原因
在这种尝试中，我需要将变量从factor传递到整数，因为我无法成功使用此类变量，例如：id1Moda$mode.num[id1Moda$Mode=="Walk"] <-1

谢谢您的帮助！

Answer 1

尝试一下。

使用联接和过滤器完成任务。内部联接对于大型数据集可能会占用大量内存

请注意，我已将finished_a的名称更改为finished_at

df1 %>% inner_join(df2, by="id_user") %>% 
  filter(tracked_at >= started_at,  tracked_at <=finished_at)


   ID          tracked_at lat long id_user          started_at         finished_at  Mode
1   2 2017-10-16 06:00:07   1    2       1 2017-10-16 06:00:05 2017-10-16 06:02:10  Walk
2   3 2017-10-16 06:01:15   1    2       1 2017-10-16 06:00:05 2017-10-16 06:02:10  Walk
3   4 2017-10-16 06:02:09   1    2       1 2017-10-16 06:00:05 2017-10-16 06:02:10  Walk
4   7 2017-10-16 06:04:15   1    2       1 2017-10-16 06:04:15 2017-10-16 06:18:30 Train
5   8 2017-10-16 06:10:15   1    2       1 2017-10-16 06:04:15 2017-10-16 06:18:30 Train
6   9 2017-10-16 06:14:15   1    2       1 2017-10-16 06:04:15 2017-10-16 06:18:30 Train
7  10 2017-10-16 06:16:15   1    2       1 2017-10-16 06:04:15 2017-10-16 06:18:30 Train
8  11 2017-10-16 06:18:30   1    2       1 2017-10-16 06:04:15 2017-10-16 06:18:30 Train
9  12 2017-10-16 06:18:45   1    2       1 2017-10-16 06:18:31 2017-10-16 06:19:45  Walk
10 13 2017-10-16 06:19:15   1    2       1 2017-10-16 06:18:31 2017-10-16 06:19:45  Walk
11 14 2017-10-16 06:19:40   1    2       1 2017-10-16 06:18:31 2017-10-16 06:19:45  Walk
12 17 2017-10-17 08:10:02   1    2       2 2017-10-17 08:10:02 2017-10-17 08:16:02   Car
13 18 2017-10-17 08:16:02   1    2       2 2017-10-17 08:10:02 2017-10-17 08:16:02   Car

中讨论了类似的问题

Join tables based on multiple ranges in R

汇总来自具有多个ifelse条件的不同长度数据集的列

1 个答案: