如何处理不规则间隔时间序列？

Question

我有一个包含各种传感器读数的CSV文件。

它看起来好像代表了四个不同的时间序列，温度，动作，功率和电池，每次读数的时间都略有不同。

dat <- read.csv(text = "id,device_id,type,value,timestamp
1432,4,temperature,21,2015-06-01T00:00:00Z
1433,4,motion,0,2015-06-01T00:00:15Z
1434,4,power,0,2015-06-01T00:00:30Z
1435,4,battery,4.16,2015-06-01T00:00:46Z
1448,4,temperature,21,2015-06-01T00:17:00Z
1449,4,motion,0,2015-06-01T00:17:15Z
1450,4,power,0,2015-06-01T00:17:30Z
1451,4,battery,4.16,2015-06-01T00:17:45Z
1464,4,temperature,21,2015-06-01T00:33:57Z")

str(dat)

'data.frame':   9 obs. of  5 variables:
 $ id       : int  1432 1433 1434 1435 1448 1449 1450 1451 1464
 $ device_id: int  4 4 4 4 4 4 4 4 4
 $ type     : Factor w/ 4 levels "battery","motion",..: 4 2 3 1 4 2 3 1 4
 $ value    : num  21 0 0 4.16 21 0 0 4.16 21
 $ timestamp: Factor w/ 9 levels "2015-06-01T00:00:00Z",..: 1 2 3 4 5 6 7 8 9

如何操纵数据以根据时间绘制四个图形？

Answer 1

只需几行代码就可以在R中轻松完成此操作。

第一步是将时间戳转换为POSIXct日期。最简单的方法是在导入期间将此列强制为类character：

dat <- read.csv(colClasses = c(timestamp = "character"),
                text = "id,device_id,type,value,timestamp
                1432,4,temperature,21,2015-06-01T00:00:00Z
                1433,4,motion,0,2015-06-01T00:00:15Z
                1434,4,power,0,2015-06-01T00:00:30Z
                1435,4,battery,4.16,2015-06-01T00:00:46Z
                1448,4,temperature,21,2015-06-01T00:17:00Z
                1449,4,motion,0,2015-06-01T00:17:15Z
                1450,4,power,0,2015-06-01T00:17:30Z
                1451,4,battery,4.16,2015-06-01T00:17:45Z
                1464,4,temperature,21,2015-06-01T00:33:57Z")

然后转换为POSIXct。有关?strptime（完整日期）和%F（全职）的含义，请参阅%T：

dat$timestamp <- as.POSIXct(dat$timestamp, format = "%FT%TZ")
str(dat)

'data.frame':   9 obs. of  5 variables:
 $ id       : int  1432 1433 1434 1435 1448 1449 1450 1451 1464
 $ device_id: int  4 4 4 4 4 4 4 4 4
 $ type     : Factor w/ 4 levels "battery","motion",..: 4 2 3 1 4 2 3 1 4
 $ value    : num  21 0 0 4.16 21 0 0 4.16 21
 $ timestamp: POSIXct, format: "2015-06-01 00:00:00" "2015-06-01 00:00:15" ...

然后使用ggplot2进行绘图。例如：

library(ggplot2)
ggplot(dat, aes(x=timestamp, y = value, group = device_id)) + 
  geom_point() + 
  geom_line() +
  facet_grid(type ~ ., scales = "free_y")