行之间的时间差乘以因子R

Question

我有这个样本：

data <- structure(list(timestamp = c(1401581040991, 1401581230769, 1401581410907, 
1401581591597, 1401581960830, 1401582002091, 1401582140958, 1401582330515, 
1401585071017, 1401585432174, 1401585641225, 1401586011911, 1401587120695, 
1401588721173, 1401589081689, 1401581041819, 1401585363131, 1401586083812, 
1401586983743, 1401588785148), timestamp_pretty = structure(c(1L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 12L, 13L, 14L, 17L, 18L, 20L, 
2L, 11L, 15L, 16L, 19L), .Label = c("01/06/2014 00:04:00", "01/06/2014 00:04:01", 
"01/06/2014 00:07:10", "01/06/2014 00:10:10", "01/06/2014 00:13:11", 
"01/06/2014 00:19:20", "01/06/2014 00:20:02", "01/06/2014 00:22:20", 
"01/06/2014 00:25:30", "01/06/2014 01:11:11", "01/06/2014 01:16:03", 
"01/06/2014 01:17:12", "01/06/2014 01:20:41", "01/06/2014 01:26:51", 
"01/06/2014 01:28:03", "01/06/2014 01:43:03", "01/06/2014 01:45:20", 
"01/06/2014 02:12:01", "01/06/2014 02:13:05", "01/06/2014 02:18:01"
), class = "factor"), mmsi = c(205477000L, 205477000L, 205477000L, 
205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 
205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 
205482000L, 205482000L, 205482000L, 205482000L, 205482000L), 
    diff_time_seconds = c(NA, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    2L, 1L, 1L, 3L, 1L, 2L, NA, 9L, 4L, 1L, 3L)), .Names = c("timestamp", 
"timestamp_pretty", "mmsi", "diff_time_seconds"), row.names = c(NA, 
-20L), class = c("data.table", "data.frame"), .internal.selfref = <pointer: 0x00000000001e0788>, sorted = "mmsi")

我希望每个因素的每一行之间都有时间差（以秒为单位）。 例如，对于一个因子的首次出现，时间差为0。

我可以用列timestamp （纪元时间）做到这一点。 但是，当我尝试使用timestamp_pretty ，它变得很疯狂。 我一直在四处寻找，但找不到解决方案，但我记得几天前就知道了...

这是我的输出示例：您可以清楚地看到diff_time_seconds_timestamp_pretty是不正确的...

structure(list(timestamp = c(1401581040991, 1401581230769, 1401581410907, 
1401581591597, 1401581960830, 1401582002091, 1401582140958, 1401582330515, 
1401585071017, 1401585432174, 1401585641225, 1401586011911, 1401587120695, 
1401588721173, 1401589081689, 1401581041819, 1401585363131, 1401586083812, 
1401586983743, 1401588785148), timestamp_pretty = structure(c(1L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 12L, 13L, 14L, 17L, 18L, 20L, 
2L, 11L, 15L, 16L, 19L), .Label = c("01/06/2014 00:04:00", "01/06/2014 00:04:01", 
"01/06/2014 00:07:10", "01/06/2014 00:10:10", "01/06/2014 00:13:11", 
"01/06/2014 00:19:20", "01/06/2014 00:20:02", "01/06/2014 00:22:20", 
"01/06/2014 00:25:30", "01/06/2014 01:11:11", "01/06/2014 01:16:03", 
"01/06/2014 01:17:12", "01/06/2014 01:20:41", "01/06/2014 01:26:51", 
"01/06/2014 01:28:03", "01/06/2014 01:43:03", "01/06/2014 01:45:20", 
"01/06/2014 02:12:01", "01/06/2014 02:13:05", "01/06/2014 02:18:01"
), class = "factor"), mmsi = c(205477000L, 205477000L, 205477000L, 
205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 
205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 205477000L, 
205482000L, 205482000L, 205482000L, 205482000L, 205482000L), 
    diff_time_seconds_timestamp_pretty = c(NA, 2L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 2L, 1L, 1L, 3L, 1L, 2L, NA, 9L, 4L, 1L, 3L
    ), diff_time_seconds_timestamp = c(NA, 189778, 180138, 180690, 
    369233, 41261, 138867, 189557, 2740502, 361157, 209051, 370686, 
    1108784, 1600478, 360516, NA, 4321312, 720681, 899931, 1801405
    )), .Names = c("timestamp", "timestamp_pretty", "mmsi", "diff_time_seconds_timestamp_pretty", 
"diff_time_seconds_timestamp"), row.names = c(NA, -20L), class = c("data.table", 
"data.frame"), .internal.selfref = <pointer: 0x00000000001e0788>, sorted = "mmsi")

我使用以下代码：

data <- data[,c("timestamp", "timestamp_pretty","mmsi")] 
data <- data[order(data$mmsi, data$timestamp_pretty),] 
library("data.table")
data<-data.table(data)
setkey(data,mmsi)
data[,diff_time_seconds_timestamp_pretty:=c(NA,diff(timestamp_pretty)),by=mmsi]
options(digits=12)
data[,diff_time_seconds_timestamp:=c(NA,diff(timestamp)),by=mmsi]

我认为diff_time_seconds_timestamp_pretty的格式有diff_time_seconds_timestamp_pretty但是我对此感到diff_time_seconds_timestamp_pretty ！

谢谢！

Answer 1

diff(timestamp_pretty)适用于因子的数值，而不适用于日期（无论如何，它不适用于字符）。

像这样转换或创建POSIXct列（我确实选择了第二种方式）：

data[, pos_time:=as.POSIXct( timestamp_pretty, format="%d/%m/%Y %H:%M:%S", tz="UTC" )]

现在，差异将起作用：

data[,diff_time_seconds_timestamp_pretty:=c(NA,diff(pos_time)),by="mmsi"]

这使：

> head(data)
      timestamp    timestamp_pretty      mmsi diff_time_seconds            pos_time diff_time_seconds_timestamp_pretty
1: 1.401581e+12 01/06/2014 00:04:00 205477000                NA 2014-06-01 00:04:00                                 NA
2: 1.401581e+12 01/06/2014 00:07:10 205477000                 2 2014-06-01 00:07:10                                190
3: 1.401581e+12 01/06/2014 00:10:10 205477000                 1 2014-06-01 00:10:10                                180
4: 1.401582e+12 01/06/2014 00:13:11 205477000                 1 2014-06-01 00:13:11                                181
5: 1.401582e+12 01/06/2014 00:19:20 205477000                 1 2014-06-01 00:19:20                                369
6: 1.401582e+12 01/06/2014 00:20:02 205477000                 1 2014-06-01 00:20:02                                 42