data.table (I): cruces

dim(qjilm)
# [1] 3218575 5
dim(tf)
# [1] 6340091 7
dim(tfe)
#[1] 1493772 3

head(qjilm, 2)
#pos.es length.en length.es pos.en qjilm
#1 1 2 1 1 0.8890203
#2 1 2 1 2 0.1109797

head(tf, 2)
#frase es pos.es length.es en pos.en length.en
#1 996 ! 42 42 ! 43 44
#2 1231 ! 37 37 ! 37 38

head(tfe, 2)
#en es tfe
#1 ! ! 4.364360e-01
#2 ! !" 4.945229e-24

res <- merge(merge(tf, tfe), qjilm)

res <- merge(merge(tf, tfe), qjilm)
#user system elapsed
#442.991 2.496 446.832

dim(res)
#[1] 6340091 9

library(data.table)

system.time({
    res.dt <- merge(data.table( tf,  key = c("en", "es")),
                    data.table( tfe, key = c("en", "es")) )

    res.dt <- merge(
        setkeyv(res.dt,
            cols = c("pos.es", "pos.en", "length.es", "length.en")),
        data.table(qjilm,
            key  = c("pos.es", "pos.en", "length.es", "length.en"))
    )
})
#user system elapsed
#32.070 0.012 32.118

dim(res.dt)
#[1] 6340091 9

tf.dt  <- data.table( tf,  key = c("en", "es"))
tfe.dt <- data.table( tfe, key = c("en", "es"))

system.time( res <- merge(tf.dt, tfe.dt) )
#user system elapsed
#3.464 0.000 3.466

dim(res)
#[1] 6340091 8