将具有多个时区的大型数据集的本地日期时间转换为 UTC

Question

我得到了一个包含纬度、经度、本地日期和本地时间字段的海量数据集。我试图将此信息组织到 ISO UTC 时间字段中。我的代码有问题，因为当数据中有多个时区时，我不知道如何使用 as.POSIXct() 。每当我尝试在 tz = 部分使用数组或循环时，我都会收到错误消息。

我使用了 lutz 包中的 tz_lookup_coords() 函数来定义数据框中每一行的时区。我还能够将我的数据过滤到单个时区，并成功使用 as.POSIXct() 和 format() 函数来获取 UTC 时间。但是，我想要一个更优雅的解决方案，将单个代码应用于整个数据集。

sample<- data.frame(
  "localDate" = c("2015-04-20","2016-07-17","2015-08-06"), 
  "localTime" = c("14:00", "14:46", NA ), 
  "timeZone" = c("Pacific/Pago_Pago", NA , "Pacific/Honolulu")
)

# Change times from local to UTC
sample$localDateTime <- paste(sample$localDate, sample$localTime, sep = " ")
for (i in 1:nrow(sample)) {
  sample[i,]$localDateTime <- as.POSIXct(sample[i,]$localDateTime, tz= sample[i,]$timeZone, "%Y-%m-%d %H:%M")
}
sample$eventDate <- format(sample$localDateTime, tz= "UTC", usetz = TRUE)

当我输入像 "Pacific/Honolulu" 这样的单个时区时，代码运行良好，但它会将每一行视为在一个时区中。

> sample
   localDate localTime          timeZone    localDateTime               eventDate
1 2015-04-20     14:00 Pacific/Pago_Pago 2015-04-20 14:00 2015-04-21 00:00:00 UTC
2 2016-07-17     14:46    Pacific/Saipan 2016-07-17 14:46 2016-07-18 00:46:00 UTC
3 2015-08-06     10:35  Pacific/Honolulu 2015-08-06 10:35 2015-08-06 20:35:00 UTC

如果我尝试在函数的 tz = 部分使用引号中的字符串以外的任何内容，我会收到此代码：

Error in strptime(x, format, tz = tz) : invalid 'tz' value

Answer 1

library(lubridate)

sample <- 
  data.frame(
     "localDate" = c("2015-04-20","2016-07-17","2015-08-06"), 
     "localTime" = c("14:00", "14:46", "10:35"), 
     "timeZone" = c("Pacific/Pago_Pago", "Pacific/Saipan", "Pacific/Honolulu")
   )

sample$localDateTime <- 
  paste(sample$localDate, sample$localTime, sep = " ")

list <- 
  list()

for (i in 1:nrow(sample)){
     list[[i]] <- 
       ymd_hm(sample$localDateTime[i], 
              tz = as.character(sample$timeZone[i]))
   }

list   

R> list
[[1]]
[1] "2015-04-20 14:00:00 SST"

[[2]]
[1] "2016-07-17 14:46:00 ChST"

[[3]]
[1] "2015-08-06 10:35:00 HST"

Answer 2

错误是由于 timeZone 列是一个因子而不是字符向量，定义 data.frame 时使用 stringsAsFactors = FALSE 指定 timeZone 为字符柱子。您还可以通过使用 lubridate 包中的向量化函数来避免任何循环：

library(lubridate)

df <- data.frame(
    "localDate" = c("2015-04-20","2016-07-17","2015-08-06"), 
    "localTime" = c("14:00", "14:46", "10:35"), 
    "timeZone" = c("Pacific/Pago_Pago", "Pacific/Saipan", "Pacific/Honolulu"),
    stringsAsFactors = FALSE
)

df$eventDate <- force_tzs(ymd_hm(with(df, paste(localDate, localTime))), tzones = df$timeZone)

df
#>    localDate localTime          timeZone           eventDate
#> 1 2015-04-20     14:00 Pacific/Pago_Pago 2015-04-21 01:00:00
#> 2 2016-07-17     14:46    Pacific/Saipan 2016-07-17 04:46:00
#> 3 2015-08-06     10:35  Pacific/Honolulu 2015-08-06 20:35:00

编辑：在缺少值的情况下，检查每一行是否可以转换，如果不能return NA。下面是一个使用 base R 的示例解决方案：

df <- data.frame(
    "localDate" = c("2015-04-20","2016-07-17","2015-08-06", "2019-01-01", "2019-01-01"), 
    "localTime" = c("14:00", "14:46", "10:35", NA, "00:00"), 
    "timeZone" = c("Pacific/Pago_Pago", "Pacific/Saipan", "Pacific/Honolulu", 
        "Pacific/Honolulu", NA),
    stringsAsFactors = FALSE
)

df$eventDate <- apply(df, 1, function(row) {     
      ifelse(any(is.na(row)), NA_character_,
          format(as.POSIXct(paste(row["localDate"], row["localTime"]), "%Y-%m-%d %H:%M",
              tz = row["timeZone"]), tz = "UTC", usetz = TRUE)
      )
    })
df   
#>    localDate localTime          timeZone               eventDate
#> 1 2015-04-20     14:00 Pacific/Pago_Pago 2015-04-21 01:00:00 UTC
#> 2 2016-07-17     14:46    Pacific/Saipan 2016-07-17 04:46:00 UTC
#> 3 2015-08-06     10:35  Pacific/Honolulu 2015-08-06 20:35:00 UTC
#> 4 2019-01-01      <NA>  Pacific/Honolulu                    <NA>
#> 5 2019-01-01     00:00              <NA>                    <NA>

将具有多个时区的大型数据集的本地日期时间转换为 UTC

Converting local date-times from a large dataset with multiple time zones into UTC

timezone

r

utc