3

This question is a follow-up to Count days per year.

I did what Dirk suggested with a huge data.frame. My commands look like this:

dateSeq <- function(df) {
  res <- seq(as.Date(df["begin"]), as.Date(df["end"]), by = "1 day")
  format(res, "%Y")
}

dataFrame$seq <- apply(dataFrame, 1, dateSeq)
dataFrame_years <- do.call("c", dataFrame[["seq"]])

rm(dataFrame)
gc()
gc()

dataFrame_tab <- table(dataFrame_years)

Now, these commands fill up my 8 GB Ram and 2 GB swap space. In the mean time my processor is bored having a processor load of maybe 15 %.

Besides, it takes ages for my computer to fulfill my "desires". Can I shift some of the work to the CPU and unburden my Ram a bit?

Community
  • 1
  • 1
speendo
  • 13,045
  • 22
  • 71
  • 107

1 Answers1

2

Indeed, the referred solution is uneccessary memory hungry. Try this:

begin <- as.POSIXlt("2007-05-20", tz = "GMT")
end <- as.POSIXlt("2010-06-13", tz = "GMT")

year <- seq(begin$year, end$year) + 1900
year.begin <- as.POSIXlt(paste(year, "01", "01", sep="-"), tz="GMT")
year.begin[1] <- begin
year.end <- as.POSIXlt(paste(year, "12", "31", sep="-"), tz="GMT")
year.end[length(year.end)] <- end
days <- as.numeric(year.end - year.begin) + 1
cbind(year, days)
Andrei
  • 2,585
  • 1
  • 14
  • 17