Bom dia, Existe a função rle() que é útil nesses casos. ### <code r> dat <- read.table(text='Horario Nivel Status "01/01/2011 00:00" 29 0 "01/01/2011 00:01" 29 0 "01/01/2011 00:02" 30 1 "01/01/2011 00:03" 31 1 "01/01/2011 00:04" 29 0 "01/01/2011 00:05" 29 0 "01/01/2011 00:06" 29 0 "01/01/2011 00:07" 29 0 "01/01/2011 00:08" 30 1 "01/01/2011 00:09" 32 1 "01/01/2011 00:10" 34 1 "01/01/2011 00:11" 35 1 "01/01/2011 00:12" 35 1 "01/01/2011 00:13" 25 0', head=T, as.is=T) dat$Horario <- as.POSIXlt(dat$Horario, format = '%d/%m/%Y %H:%M') head(dat) res <- do.call(data.frame, rle(dat$Status)) n <- nrow(res) res$ini <- c(1, cumsum(res$lengths[-n])+1) res$fim <- cumsum(res$lengths) res$timeDiff <- with(res, dat[fim,1]-dat[ini,1]) res # lengths values ini fim timeDiff # 1 2 0 1 2 60 secs # 2 2 1 3 4 60 secs # 3 4 0 5 8 180 secs # 4 5 1 9 13 240 secs # 5 1 0 14 14 0 secs ### encontrar os intervalos de tempo em que a variavel "Status" permanence no valor 1. res[res$values==1,] # lengths values ini fim timeDiff # 2 2 1 3 4 60 secs # 4 5 1 9 13 240 secs ### </code> Éder Comunello <c <comunello.eder@gmail.com>omunello.eder@gmail.com> Dourados, MS - [22 16.5'S, 54 49'W]