In Pajek time points are represented by integers. In this note we describe how to convert dates into integers using R.
Besides the basic support for date and time ( help(“Date”) and help(“POSIXt”) ) R has also a package lubridate ( paper/PDF, CRAN ) for more user friendly work with dates and times.
For constructing Pajek network files the basic functions are enough.
> (da <- as.Date(ISOdate(2017,9,14)))
[1] "2017-09-14"
> as.numeric(da)
[1] 17423
> (db <- as.Date(as.character(20170914),"%Y%m%d"))
[1] "2017-09-14"
> as.numeric(db)
[1] 17423
> (dc <- as.Date("14.9.17","%d.%m.%y"))
[1] "2017-09-14"
> (de <- as.Date("14 Sep 2017","%d %b %Y"))
[1] "2017-09-14"
> (df <- as.Date(17423,origin="1970-01-01"))
[1] "2017-09-14"
For different options for specifying a date format see R wikibook.
The default value of the origin of dates is “1970-01-01”. Assume that we would like to use in our Pajek network description a granularity by days with the origin January 1, 2017.
We first compute the index of January 1, 2017
> (d0 <- as.Date(ISOdate(2016,12,31))) [1] "2016-12-31" > as.numeric(d0) [1] 17166
to be used in the function nDays(d) where d is a Date. The function nDays(d) returns the number of dates + 1 from January 1, 2017.
Another function nday2date(n) returns the Date corresponding to n-th day from January 1, 2017.
> nDays <- function(d) return(as.numeric(d)-17166) > nday2date <- function(n) return(as.Date(n,origin="2016-12-31")) > d <- as.Date(ISOdate(2017,9,14)) > nDays(d) [1] 257 > (n <- nDays(d)) [1] 257 > nday2date(n) [1] "2017-09-14" > d <- as.Date(ISOdate(2017,1,1)) > (n <- nDays(d)) [1] 1 > nday2date(n) [1] "2017-01-01"
Subject [Pajek] Pajek Temporal Netowk From Mohsin Adalat To pajek@list.fmf.uni-lj.si Date 2017/09/17 08:40
Hello Sir,
I have edge list like this. i want to create temporal network based on date. i have used different tools like txt2pajek3. However when i convert network into temporal, it gives error that vertex one does not have valid time record. Sir i want to create dynamic network based on dates of arcs. Is there any tool available to convert this type of data into temporal network.
a b 10-10-2016 b c 20-10-2016 d e 30-10-2016 a b 30-10-2016
Thank You.
Assume that we have a file dates.csv
from;to;date a;b;10-10-2016 b;c;20-10-2016 d;e;30-10-2016 a;b;30-10-2016
using a short program in R (set your starting date in d0 and your directory in setwd() )
> d0 <- as.numeric(as.Date(ISOdate(2016,10,1)))-1
> nDays <- function(d) return(as.numeric(d)-d0)
> setwd("C:/Users/batagelj/***/Rnet/pajek")
> links <- read.csv2("dates.csv",encoding='UTF-8',colClasses="character")
> L <- levels(factor(c(links$from,links$to))); n <- length(L)
> F <- as.numeric(factor(links$from,levels=L))
> T <- as.numeric(factor(links$to,levels=L))
> net <- file("dates.net","w"); cat('*vertices ',n,'\n',file=net)
> for(v in 1:n) cat(v,' "',L[v],'"\n',sep='',file=net)
> cat('*arcs\n',file=net)
> for(a in 1:nrow(links)) cat(F[a],' ',T[a],' 1 [',
+ nDays(as.Date(links$date[a],"%d-%m-%Y")),']\n',sep='',file=net)
> close(net)
we get a Pajek's network file dates.net
*vertices 5 1 "a" 2 "b" 3 "c" 4 "d" 5 "e" *arcs 1 2 1 [10] 2 3 1 [20] 4 5 1 [30] 1 2 1 [30]