time - R - Split numeric vector into intervals -

i have question regarding "splitting" of vector, although different approaches might feasible. have data.frame(df) looks (simplified version):

   case time 1   1   5 2   2   3 3   3   4 

the "time" variable counts units of time (days, weeks etc) until event occurs. expand data set increasing number of rows , "split" "time" intervals of length 1, beginning @ 2. result might this:

    case    time    begin   end 1   1       5       2       3 2   1       5       3       4 3   1       5       4       5 4   2       3       2       3 5   3       4       2       3 6   3       4       3       4 

obviously, data set bit larger example. feasible method achieve result?

i had 1 idea of beginning with

df.exp <- df[rep(row.names(df), df$time - 2), 1:2] 

in order expand number of rows per case, according number of time intervals. based on this, "begin" , "end" column might added in fashion of:

df.exp$begin <- 2:(df.exp$time-1) 

however, i'm not successful @ creating respective columns, because command uses first row calculate (df.exp$time-1) , doesn't automatically distinguish "case".

any ideas appreciated!

you can try

df2 <- df1[rep(1:nrow(df1), df1$time-2),] row.names(df2) <- null m1 <- do.call(rbind,           map(function(x,y) {                   v1 <- seq(x,y)                   cbind(v1[-length(v1)],v1[-1l])},                   2, df1$time)) df2[c('begin', 'end')] <- m1 df2 #  case time begin end #1    1    5     2   3 #2    1    5     3   4 #3    1    5     4   5 #4    2    3     2   3 #5    3    4     2   3 #6    3    4     3   4 

or option data.table

library(data.table) setdt(df1)[,{tmp <- seq(2, time)                list(time= time,                     begin= tmp[-length(tmp)],                     end=tmp[-1])} , = case] #   case time begin end #1:    1    5     2   3 #2:    1    5     3   4 #3:    1    5     4   5 #4:    2    3     2   3 #5:    3    4     2   3 #6:    3    4     3   4