plyr to dplyr

plyr_dplyr_test.R

df <- data.frame(id = seq(1, 500), 
               attr1 = rep(c("up", "up", "down", "down")), 
               attr2 = rep(c("left", "right", "left", "right")), 
               attr3 = rep(c("b", "a", "select", "start")),
               attr4 = sample(c("c", "o", "n", "t", "r", "a"), 100, replace=T),
               x = rnorm(100))

dfChar = df
dfChar[, names(dfChar)] <- lapply(dfChar[, names(dfChar)], as.character)
dt = as.data.table(df)
dtChar = as.data.table(dfChar)

llply(list(df,dfChar, dt, dtChar), str)

system.time(ddply(df, .(attr4), plyr::mutate, l = seq(1, length(attr4))))
system.time(ddply(df, .(attr4), dplyr::mutate, l = seq(1, length(attr4))))

system.time(ddply(dfChar, .(attr4), plyr::mutate, l = seq(1, length(attr4))))
system.time(ddply(dfChar, .(attr4), dplyr::mutate, l = seq(1, length(attr4))))

system.time(ddply(dt, .(attr4), plyr::mutate, l = seq(1, length(attr4))))
system.time(ddply(dt, .(attr4), dplyr::mutate, l = seq(1, length(attr4))))

system.time(ddply(dtChar, .(attr4), plyr::mutate, l = seq(1, length(attr4))))
system.time(ddply(dtChar, .(attr4), dplyr::mutate, l = seq(1, length(attr4))))

system.time(dt %>% group_by(attr4) %>% mutate(l = seq(1, length(x))))

system.time(a <- ddply(dt, .(attr4), dplyr::mutate, l = seq(1, length(attr4))))
system.time(b <- dt %>% group_by(attr4) %>% mutate(l = seq(1, length(x))) %>% arrange(attr4))