[R转换数据帧手动timeserie

Question

我有这样的结构中的R

DATA = read.csv("Anomaly.csv") 
DATA 

    col1  col2 
57.17339 0.00064822 
52.90232 0.00064386 
46.20606 0.00064144 
61.23981 0.00064216 

我想这样

col1  col2  col1-1 col2-1 col1-2 col2-2 
57.17339 0.00064822 NA  NA  NA  NA 
52.90232 0.00064386 57.17339 0.00064822 NA  NA 
46.20606 0.00064144 52.90232 0.00064386 57.17339 0.00064822 
61.23981 0.00064216 46.20606 0.00064144 52.90232 0.00064386 

Answer 1

回报我们可以从data.table与shift尝试。将'data.frame'转换为'data.table'（setDT(DATA)），循环访问Data.table的子集（lapply(.SD, ..）), use the shift specifying the n as 0:2. By default the shift has type` as'lag'option。

library(data.table) 
dt <- setDT(DATA)[, unlist(lapply(.SD, shift, n = 0:2), recursive=FALSE)] 

如果我们需要订购的列

dt[, order(sub("\\D+\\d", "", colnames(dt))), with = FALSE] 
#  col11  col21 col12  col22 col13  col23 
#1: 57.17339 0.00064822  NA   NA  NA   NA 
#2: 52.90232 0.00064386 57.17339 0.00064822  NA   NA 
#3: 46.20606 0.00064144 52.90232 0.00064386 57.17339 0.00064822 
#4: 61.23981 0.00064216 46.20606 0.00064144 52.90232 0.00064386 

---

或者也可以直接应用于Data.table的子集shift。

setDT(DATA)[, shift(.SD, n = 0:2)]