在ddply中将ifelse与transform一起使用 [英] Using ifelse with transform in ddply
问题描述
我正在尝试将ddply与transform结合使用，以在数据帧中使用变量ID和Date填充新变量(summary_Date).变量的值是根据要使用ifelse评估的块的长度来选择的:
I am trying to use ddply with transform to populate a new variable (summary_Date) in a dataframe with variables ID and Date. The value of the variable is chosen based on the length of the piece that is being evaluated using ifelse:
如果给定月份中ID的观察少于五个，我希望通过将日期四舍五入到最近的月份(使用软件包lubridate中的round_date)来计算summary_Date；如果给定月份中ID的观察值超过五个，则我希望summary_Date只是Date.
If there are less than five observations for an ID in a given month, I want to have summary_Date be calculated by rounding the date to the nearest month (using round_date from package lubridate); if there are more than five observations for an ID in a given month, I want the summary_Date to simply be Date.
require(plyr)
require(lubridate)
test.df <- structure(
list(ID = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,1, 1, 1, 1, 1, 1, 1, 1, 1
, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,2, 2, 2, 2, 2, 2, 2, 2)
, Date = structure(c(-247320000, -246196800, -245073600, -243864000
, -242654400, -241444800, -126273600, -123595200
, -121176000, -118497600, 1359385200, 1359388800
, 1359392400, 1359396000, 1359399600, 1359403200
, 1359406800, 1359410400, 1359414000, 1359417600
, 55598400, 56116800, 58881600, 62078400, 64756800
, 67348800, 69854400, 72964800, 76161600, 79012800
, 1358589600, 1358676000, 1358762400, 1358848800
, 1358935200, 1359021600, 1359108000, 1359194400
, 1359280800, 1359367200), tzone = "GMT"
, class = c("POSIXct", "POSIXt"))
, Val=rnorm(40))
, .Names = c("ID", "Date", "Val"), row.names = c(NA, 40L)
, class = "data.frame")
test.df <- ddply(test.df, .(ID, floor_date(Date, "month")), transform
, summary_Date=as.POSIXct(ifelse(length(ID)<5
, round_date(Date, "month")
,Date)
, origin="1970-01-01 00:00.00"
, tz="GMT")
# Included length_x to easily see the length of the subset
, length_x = length(ID))
head(test.df,5)
# floor_date(Date, "month") ID Date Val summary_Date length_x
# 1 1962-03-01 1 1962-03-01 12:00:00 -0.1037988 1962-03-01 3
# 2 1962-03-01 1 1962-03-14 12:00:00 0.2923056 1962-03-01 3
# 3 1962-03-01 1 1962-03-27 12:00:00 0.4435410 1962-03-01 3
# 4 1962-04-01 1 1962-04-10 12:00:00 0.1159164 1962-04-01 2
# 5 1962-04-01 1 1962-04-24 12:00:00 2.9824075 1962-04-01 2
ifelse语句似乎正在运行，但是"summary_Date"中的值似乎是为正在处理转换的子集计算的第一个值，而不是特定于行的值.例如，在第3行中，summary_Date应该为1962-04-01，因为日期1962-03-27 12:00:00'应该四舍五入(因为子集中的行数少于5)，而是summary_Date的第一个计算值(
The ifelse statement seems to be working, but the value in 'summary_Date' seems to be the first value calculated for the subset that transform is working on, rather than the row-specific value. For example in row 3, summary_Date should be 1962-04-01 because the date 1962-03-27 12:00:00' should be rounded up (because there are fewer than five rows in the subset), but instead the first calculated value of summary_Date (1962-03-01) is repeated in all rows in that subset.
编辑:里卡多的答案激发了我的灵感，他使用data.table在ddply中分两步进行尝试.它也可以工作:
I was inspired by Ricardo's answer using data.table to try it in two steps with ddply. It works also:
test.df <- ddply(test.df, .(ID, floor_date(Date, "month")), transform
, length_x = length(ID))
test.df <- ddply(test.df, .(ID, floor_date(Date, "month")), transform
, summary_Date=as.POSIXct(ifelse(length_x<5
, round_date(Date, "month")
,Date)
, origin="1970-01-01 00:00.00"
, tz="GMT"))
head(test.df,5)[c(1,3:7)]
# floor_date(Date, "month") ID Date Val length_x summary_Date
# 1 1962-03-01 1 1962-03-01 12:00:00 -0.1711212 3 1962-03-01
# 2 1962-03-01 1 1962-03-14 12:00:00 -0.1531571 3 1962-03-01
# 3 1962-03-01 1 1962-03-27 12:00:00 0.1256238 3 1962-04-01
# 4 1962-04-01 1 1962-04-10 12:00:00 1.4481225 2 1962-04-01
# 5 1962-04-01 1 1962-04-24 12:00:00 -0.6508731 2 1962-05-01
推荐答案
一步式ddply解决方案(也已发布为评论)
One Step ddply solution (also posted as comment)
ddply(test.df, .(ID, floor_date(Date, "month")), mutate,
length_x = length(ID),
summary_Date=as.POSIXct(ifelse(length_x < 5, round_date(Date, "month") ,Date)
, origin="1970-01-01 00:00.00", tz="GMT")
)
这篇关于在ddply中将ifelse与transform一起使用的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
