Question

I am sorry for the cryptic title but I didn’t know how to adequately summarise my problem. So here’s my question. I have a data frame with dates and a name for several entities:

 df <- data.frame(
       time=rep(as.Date(seq(as.Date("2004/1/1"), as.Date("2005/12/1"), by = "1 month ")),2),
       name=c(rep("a",24),rep("b",24))
  )
  str(df)

  'data.frame':   48 obs. of  2 variables:
   $ time: Date, format: "2004-01-01" "2004-02-01" ...
   $ name: Factor w/ 2 levels "a","b": 1 1 1 1 1 1 1 1 1 1 ...

And I have another dataframe with several unevenly spaced events:

   events <- data.frame(
          time = c("2004-12-1", "2005-8-1", "2005-6-1", "2004-4-1"),
          event = c("normal", "extraordinary", "normal", "extraordinary"),
          name = c("a", "a", "b", "b")
   )

I want to merge these two data frames in a way that the event is assigned from the either the beginning of the data set up to the event or starting with the last event up to the next event or the end of the data set. This would look something like:

    date      name  event
   2004-01-01  a     normal
   2004-01-02  a     normal 
      ...
   2004-12-01  a     extraordinary
   2005-01-01  a     extraordinary

Is there an easy way doing this in R that I don’t see or do I merge these by hand? Thank you very much for your help!

Answer 1

I don’t know any function to do this, but here is some R code to do it yourself :

# Needed type coercions (Date for comparisons, characters to avoid 'factor' problems)
events$time <- as.Date(events$time)
events$event <- as.character(events$event)
events$name <- as.character(events$name)
df$name <- as.character(df$name)

# Events ordering (needed to detect previous events as non NA)
events <- events[ order(events$time) ,]

# Updates
df$event = NA
for(i in 1:nrow(events)) {
    # Update where time is lesser than the limit, if names correspond and if an event was not already assigned to the row
    df[ df$time <= events[i,"time"] & df$name == events[i,"name"] & is.na(df$event) , "event" ] = events[i,"event"]
}