在 R 中合并 xts - 将字符转换为 NA答案

【问题标题】：Merging xts in R - Converting Characters to NA在 R 中合并 xts - 将字符转换为 NA
【发布时间】：2019-02-25 18:21:58
【问题描述】：

我有 3 个 xts 对象

logged <- xts::xts(x = loggedInUsers$loggedInUsers, order.by = Sys.time())
loadValue <- xts::xts(x = loadAvg, order.by = Sys.time())
hostname <- xts::xts(x = loadHost, order.by = Sys.time())

dput(hostname)
dput(loadValue)
dput(logged)

dput 给出以下结果

 structure("deliverforgoodportal", .Dim = c(1L, 1L), index = structure(1551088127.27724, tzone = "", tclass = c("POSIXct",
    "POSIXt")), class = c("xts", "zoo"), .indexCLASS = c("POSIXct",
    "POSIXt"), tclass = c("POSIXct", "POSIXt"), .indexTZ = "", tzone = "")

structure(0, .Dim = c(1L, 1L), .Dimnames = list(NULL, "load"), index = structure(1551088127.27676, tzone = "", tclass = c("POSIXct",
"POSIXt")), .indexCLASS = c("POSIXct", "POSIXt"), tclass = c("POSIXct",
"POSIXt"), .indexTZ = "", tzone = "", class = c("xts", "zoo"))

structure(1, .Dim = c(1L, 1L), index = structure(1551088127.27637, tzone = "", tclass = c("POSIXct",
"POSIXt")), class = c("xts", "zoo"), .indexCLASS = c("POSIXct",
"POSIXt"), tclass = c("POSIXct", "POSIXt"), .indexTZ = "", tzone = "")

当我合并这三个并打印时，主机名被转换为 NA

  tmp <- merge.xts(hostname, logged, loadValue, all = TRUE)
    print(tmp)

输出是：（主机名是NA）

                    hostname logged  load
2019-02-25 09:48:47       NA      1    NA
2019-02-25 09:48:47       NA     NA    0
2019-02-25 09:48:47       NA     NA    NA

为什么会以 NA 的形式出现？

【问题讨论】：

标签： r xts

【解决方案1】：

您应该意识到 xts 对象是一个时间序列和一个矩阵。现在矩阵只能包含一种类型的值，字符或数字。但不是两者兼而有之。您的合并试图将字符值矩阵（主机名）与数值（记录和加载）结合起来。这会导致主机名值被强制为 NA。

如果要加入这些数据，则必须使用 data.frame（或 data.table）。另请注意，您的时间值不相等，它们以毫秒为单位。因此，如果您想在几分钟内加入，请首先使用 lubridate 包中的floor_date。见下面两个有和没有润滑的例子。我使用包 timetk 将 xts 对象转换为 tibble，但取决于您可能不需要的源数据。

使用 full_join，无润滑

library(timetk)
library(dplyr)
hostname <- tk_tbl(hostname)
loadValue <- tk_tbl(loadValue)
logged <- tk_tbl(logged)

hostname %>% 
  full_join(loadValue) %>% 
  full_join(logged, 
            by = "index", 
            suffix = c("_hostname", "_logged"))

Joining, by = "index"
# A tibble: 3 x 4
  index               value_hostname        load value_logged
  <dttm>              <chr>                <dbl>        <dbl>
1 2019-02-25 10:48:47 deliverforgoodportal    NA           NA
2 2019-02-25 10:48:47 NA                       0           NA
3 2019-02-25 10:48:47 NA                      NA            1

使用 lubridate 和左连接：

hostname %>% 
  mutate(index = lubridate::floor_date(index, unit = "seconds")) %>% 
  left_join(loadValue %>% mutate(index = lubridate::floor_date(index, unit = "seconds"))) %>% 
  left_join(logged %>% mutate(index = lubridate::floor_date(index, unit = "seconds")), 
            by = "index", 
            suffix = c("_hostname", "_logged"))    

Joining, by = "index"
# A tibble: 1 x 4
  index               value_hostname        load value_logged
  <dttm>              <chr>                <dbl>        <dbl>
1 2019-02-25 10:48:47 deliverforgoodportal     0            1

【讨论】：