问题描述
我用 R 构建了一个数据框，从我的电子邮件中读取了一些事件。 基本上，最终数据帧的结构如下：
'data.frame': 74 obs. of 7 variables:
$ process_name : Factor w/ 2 levels : 1 1 1 1 2 2 2 2 2 2 ...
$ job_code : chr "TRB1619825404" "TRB1619825404" "TRB1619825404" "TRB1619825404" ...
$ phase : Factor w/ 7 levels,..: 4 4 6 6 4 5 7 1 3 2 ...
$ stage : Factor w/ 2 levels "End","Start": 2 1 2 1 2 1 2 2 2 2 ...
$ date : POSIXct,format: "2021-04-30 23:30:04" "2021-05-01 01:57:26" "2021-05-01 01:57:26" "2021-05-01 02:25:26" ...
$ execution_date: Date,format: "2021-04-30" "2021-05-01" "2021-05-01" "2021-05-01" ...
$ execution_time : 'hms' num 23:30:04 01:57:26 01:57:26 02:25:26 ...
..- attr(*,"units")= chr "secs"
每个事件都有一个相关的开始和结束日期和时间。 我想要做的（为了计算事件的持续时间）是将数据帧转换成这样的：
'data.frame'
$ process_name
$ job_code
$ phase
$ start_date
$ end_date
$ duration
我尝试使用 dcast，但它使用默认聚合函数，我只想重塑数据帧。 有什么想法吗？
解决方法
首先，我猜你的数据是这样的：
dat <- structure(list(process_name = structure(c(1L,1L,1L),.Label = c("L01","L02"),class = "factor"),job_code = c("TRB1619825404","TRB1619825404","TRB1619825404"),phase = structure(c(3L,3L,5L,5L),.Label = c("L02","L03","L04","L05","L06","L07"),stage = structure(c(2L,2L,.Label = c("End","Start" ),date = structure(c(1619825404,1619834246,1619835926),class = c("POSIXct","POSIXt"),tzone = "UTC"),execution_date = structure(c(18747,18748,18748),class = "Date"),execution_time = c("30:04","01:57:26","02:25:26" )),row.names = c(NA,4L),class = "data.frame")
（如果没有，请以我们可以使用的方式提供您的数据。）
从这里：
out <- reshape2::dcast(dat[,1:5],process_name + job_code + phase ~ stage,value.var = "date")
out[,c("End","Start")] <- lapply(out[,"Start")],as.POSIXct,origin = "1970-01-01",tz = "UTC")
out$duration <- difftime(out$End,out$Start,units = "mins")
out
# process_name job_code phase End Start duration
# 1 L01 TRB1619825404 L04 2021-05-01 01:57:26 2021-04-30 23:30:04 147.3667 mins
# 2 L01 TRB1619825404 L06 2021-05-01 02:25:26 2021-05-01 01:57:26 28.0000 mins
我不得不重新as.POSIXct，因为dcast将它们强制转换为numeric（不知道如何解决）。