CODEWITHSUNDEEP
rtime
Rahul Kumar Vij
Rahul Kumar Vij

Reputation: 83

convert a variable into timeseries data in R

I have a vector in r say

a = log10(1:1000) + rnorm(1000)

I want to convert this data set based on a parameter given say

n = 10

A function say

b = function(a,x){...........................}

d = b(a,x)

such that my data should be converted to

d = 
a1   a2   a3   a4   a5   a6   a7   a8   a9   a10
a2   a3   a4   a5   a6   a7   a8   a9   a10  a11 
..................................
.................................
a990 a991 a992 a993 a994 a995 a996 a997 a998 a999
a991 a992 a993 a994 a995 a996 a997 a998 a999 a1000

Upvotes: 0

Views: 45

Answers (3)

thelatemail
thelatemail

Reputation: 93803

A bit simpler using ?embed:

embed(a,n)[,n:1]

It doesn't have the same names, and returns a matrix, but otherwise it is identical to your longer result:

identical( embed(a,n)[,n:1], unname(as.matrix(b)) )
#[1] TRUE

Upvotes: 1

tushaR
tushaR

Reputation: 3116

Try this:

b= function(a,n){
    lst=sapply(1:(length(a)-n),function(t,a,n){a[t:(n+t-1)]},a,n,simplify = F)
    Reduce(function(x,y){rbind(x,y)},lst)
    }

Upvotes: 1

Rahul Kumar Vij
Rahul Kumar Vij

Reputation: 83

a = log10(1:1000)+rnorm(1000,0,2)
n = 10
b = matrix(rep(a,n),nrow=length(a))
for(i in 1:n){
  b[1:length(a),i] = c(rep(NA,n-i),b[1:(length(a)-n+i),i])
}

b = na.omit(b)

b = as.data.frame(b)

Upvotes: 0

Related Questions