Subset and order

Hi,
You could also try ?data.table()
x<- read.table(text="a??? b??? c
1??? 2??? 3
3??? 3??? 4
2??? 4??? 5
1??? 3??? 4
",sep="",header=TRUE)

library(data.table)

xt<- data.table(xt)
?setkey(xt,a)
?subset(xt,b==3)
#?? a b c
#1: 1 3 4
#2: 3 3 4

?iord <- order(x$a)
?subset(x[iord, ], b == 3) 
#? a b c
#4 1 3 4
#2 3 3 4

Speed comparison:
set.seed(12345)
dat1<- as.data.frame(matrix(sample(1:10,3*1e7,replace=TRUE),ncol=3))
colnames(dat1)<-letters[1:3]
system.time({
iord <- order(dat1$a)
res1<-subset(dat1[iord, ], b == 3)
})
#? user? system elapsed 
#? 6.888?? 0.296?? 7.202 

dt1<- data.table(dat1)
system.time({setkey(dt1,a)
??? resdt1<-subset(dt1,b==3)})
# user? system elapsed 
#?? 0.72??? 0.06??? 0.78?

head(resdt1)
#?? a b? c
#1: 1 3? 6
#2: 1 3? 4
#3: 1 3 10
#4: 1 3? 2
#5: 1 3? 9
#6: 1 3? 8
?head(res1)
#??? a b? c
#75? 1 3? 6
#93? 1 3? 4
#300 1 3 10
#301 1 3? 2
#437 1 3? 9
#672 1 3? 8

A.K.
----- Original Message -----
From: Rui Barradas <ruipbarradas at sapo.pt>
To: Noah Silverman <noahsilverman at ucla.edu>
Cc: "R-help at r-project.org" <r-help at r-project.org>
Sent: Friday, July 5, 2013 3:51 PM
Subject: Re: [R] Subset and order

Hello,

If time is one of the problems, precompute an ordered index, and use it 
every time you want the df sorted. But that would mean you can't do it 
in a single operation.

iord <- order(x$a)
subset(x[iord, ], b == 3)

Rui Barradas

Em 05-07-2013 20:47, Noah Silverman escreveu:

Subset and order

Thread (7 messages)