Question

在尝试随时间循环时，我在R中遇到错误。这是我的数据框的子集（包含120000行）。

                 time value      mean group
1 2017-01-01 12:00:00 0.507 0.5106533    NA
2 2017-01-01 12:05:00 0.526 0.5106533    NA
3 2017-01-01 12:10:00 0.489 0.5106533    NA
4 2017-01-01 12:15:00 0.598 0.5106533    NA
5 2017-01-01 12:20:00 0.564 0.5106533    NA
6 2017-01-01 12:25:00 0.536 0.5106533    NA

让我们说我想根据时间段创建组，并获得这样的预期结果：

                 time value      mean group
1 2017-01-01 12:00:00 0.507 0.5106533    A
2 2017-01-01 12:05:00 0.526 0.5106533    A
3 2017-01-01 12:10:00 0.489 0.5106533    B
4 2017-01-01 12:15:00 0.598 0.5106533    B
5 2017-01-01 12:20:00 0.564 0.5106533    C
6 2017-01-01 12:25:00 0.536 0.5106533    C

我尝试了以下代码：

for (i in 1:length(merged.data$group)){
  if (merged.data[as.POSIXlt(i)$time >= "2017-05-15 12:00:00 GMT" & 
as.POSIXlt(i)$time <= "2017-05-29 12:00:00 GMT",]){
   merged.data$group == "A"} 
  else if (merged.data[as.POSIXlt(i)$time >= "2017-08-11 12:00:00" & 
as.POSIXlt(i)$time <= "2017-11-29 16:00:00",]){
    merged.data$group == "B"}
  else if (merged.data[as.POSIXlt(i)$time >= "2018-01-05 12:00:00" & 
as.POSIXlt(i)$time <= "2018-02-16 16:00:00",]){
    merged.data$group == "C"}
}

我收到以下错误：

Error in as.POSIXlt.numeric(i) : 'origin' must be supplied

我不明白，我认为POSIXlt摆脱了起源问题？虽然，我承认我对R中的时间问题的理解有些混乱，并且每次需要处理时间/日期时，我都会遇到一些困难的编码...

因此，我希望有人可以帮助我，不要犹豫告诉我我是否不清楚，或者是否需要更多/更好的信息来回答我的问题。

感谢您提前stackoverflowers！

Answer 1

data.table方法...

样本数据

library( data.table )

dt <- fread("time value mean 
2017-01-01T12:00:00 0.507 0.5106533    
2017-01-01T12:05:00 0.526 0.5106533    
2017-01-01T12:10:00 0.489 0.5106533   
2017-01-01T12:15:00 0.598 0.5106533    
2017-01-01T12:20:00 0.564 0.5106533    
2017-01-01T12:25:00 0.536 0.5106533    ", header = TRUE)

dt[, time := as.POSIXct( time, format = "%Y-%m-%dT%H:%M:%S" )]

代码

library( data.table )
library( lubridate )

dt[, group := LETTERS[.GRP], by = lubridate::floor_date( time, "10 mins" ) ]

#             time value      mean group
# 1: 2017-01-01 12:00:00 0.507 0.5106533     A
# 2: 2017-01-01 12:05:00 0.526 0.5106533     A
# 3: 2017-01-01 12:10:00 0.489 0.5106533     B
# 4: 2017-01-01 12:15:00 0.598 0.5106533     B
# 5: 2017-01-01 12:20:00 0.564 0.5106533     C
# 6: 2017-01-01 12:25:00 0.536 0.5106533     C

更新

根据提供的示例数据和代码，使用foverlaps进行访问

library( data.table )

#create lookup-table with periods and group-names  
periods.dt <- data.table( 
  start = as.POSIXct( c( "2017-05-15 12:00:00", "2017-08-11 12:00:00", "2018-01-05 12:00:00" ), tz = "GMT" ),
  stop = as.POSIXct( c( "2017-08-11 12:00:00", "2018-01-05 12:00:00", "2018-02-16 16:00:00"), tz = "GMT" ),
  group = LETTERS[1:3] )
#set keys
setkey( periods.dt, start, stop ) 

#create sample data
dt <- fread("time value mean 
            2017-01-01T12:00:00 0.507 0.5106533    
            2017-01-01T12:05:00 0.526 0.5106533    
            2017-01-01T12:10:00 0.489 0.5106533   
            2017-01-01T12:15:00 0.598 0.5106533    
            2017-01-01T12:20:00 0.564 0.5106533    
            2017-01-01T12:25:00 0.536 0.5106533    ", header = TRUE)

dt[, time := as.POSIXct( time, format = "%Y-%m-%dT%H:%M:%S", tz = "GMT" )]

#create dummies to join on
dt[, `:=`( start = time, stop = time )]

#perform overlap join, no match --> NA
foverlaps( dt, periods.dt, type = "within", nomatch = NA)[, c("time", "value","mean","group"), with = FALSE]
#                   time value      mean group
# 1: 2017-01-01 12:00:00 0.507 0.5106533  <NA>
# 2: 2017-01-01 12:05:00 0.526 0.5106533  <NA>
# 3: 2017-01-01 12:10:00 0.489 0.5106533  <NA>
# 4: 2017-01-01 12:15:00 0.598 0.5106533  <NA>
# 5: 2017-01-01 12:20:00 0.564 0.5106533  <NA>
# 6: 2017-01-01 12:25:00 0.536 0.5106533  <NA>

Answer 2

感谢您的回答，我发现只有日期对我有用，因为我的数据集有很大的空白。通过简单的ifelse，我发现了一些可行的方法：

merged.data $ group <-ifelse（merged.data $ date> =“ 2017-05-15”＆merged.data $ date <=“ 2017-05-29”，1， ifelse（merged.data $ date> =“ 2017-08-11”＆merged.data $ date <=“ 2017-11-29”，2， ifelse（merged.data $ date> =“ 2018-01-05”＆merged.data $ date <=“ 2018-02-16”，3，不适用）））

这不适用于我拥有的POSIXlt对象，但是Wimpel提供的解决方案似乎可行（我在使用data.table时遇到问题，但这又是另一回事了！）

再次感谢您，这个论坛确实有很大帮助！

R中posixlt之间的循环

2 个答案:

更新