R help - May 2010 - Create counter variable for subsets without a loop

Hi all,

I am looking to create a rank variable based on a continuous variable
for subsets of the data. For example, for an R integrated data set
about US states this is how a loop could create what I want:

### Example with loop
data <- cbind(state.region,as.data.frame(state.x77))[,1:2]     #
choosing a subset of the data
data <- data[order(data$state.region, 1/data$Population),]    #
ordering the data
regions <- levels(data$state.region)
temp <- NULL
ranks <- NULL
for (i in 1:length(regions)){
    temp <-
rev(rank(data[data$state.region==regions[i],"Population"]))
    ranks <- c(ranks,temp)
  }
data$rank <- ranks
data

where data$rank is the rank of the state by population within a region.

However, using loops is slow and cumbersome. I have a fairly large
data set with many subgroups and the loop runs a long time. Can
someone suggest a way to create such rank variable for subsets without
using a loop?

Thank you,
Thomas

take a look at the by, ave, aggregate and apply functions, perhaps one suits
your needs

Bart
-- 
View this message in context:
http://r.789695.n4.nabble.com/Create-counter-variable-for-subsets-without-a-loop-tp2220663p2220925.html
Sent from the R help mailing list archive at Nabble.com.

Solved it another way, without apply:


data2 <- data[order(data$state.region,-data$Population),]
cx <- as.numeric(data2$state.region)
data2$rank <- cumsum(rep(1,length(cx)))-match(cx,cx) + 1
all.equal(data2==data)


Bart
-- 
View this message in context:
http://r.789695.n4.nabble.com/Create-counter-variable-for-subsets-without-a-loop-tp2220663p2221052.html
Sent from the R help mailing list archive at Nabble.com.

Here are four solutions:

data <- cbind(state.region,as.data.frame(state.x77))[,1:2]

# ave
data2 <- data[order(data$state.region, -data$Population), ]
data2$rank <- ave(data2$Population, data2$state.region, FUN = seq_len))

# by
f <- function(x) cbind(x[order(-x$Population), ], rank = 1:nrow(x))
do.call("rbind", by(data, data$state.region, f))

# ddply - same f as in by solution
library(plyr)
ddply(data, .(state.region), f)

# sqldf with PostgreSQL
library(RpgSQL)
library(sqldf)
sqldf('select
   *, rank() over (partition by "state.region" order by
"Population" desc)
   from data
   order by "state.region", "Population" desc')


On Mon, May 17, 2010 at 5:32 PM, Thomas Brambor <tbrambor at stanford.edu>
wrote:
> Hi all,
>
> I am looking to create a rank variable based on a continuous variable
> for subsets of the data. For example, for an R integrated data set
> about US states this is how a loop could create what I want:
>
> ### Example with loop
> data <- cbind(state.region,as.data.frame(state.x77))[,1:2] ? ? #
> choosing a subset of the data
> data <- data[order(data$state.region, 1/data$Population),] ? ?#
> ordering the data
> regions <- levels(data$state.region)
> temp <- NULL
> ranks <- NULL
> for (i in 1:length(regions)){
> ? ?temp <-
rev(rank(data[data$state.region==regions[i],"Population"]))
> ? ?ranks <- c(ranks,temp)
> ?}
> data$rank <- ranks
> data
>
> where data$rank is the rank of the state by population within a region.
>
> However, using loops is slow and cumbersome. I have a fairly large
> data set with many subgroups and the loop runs a long time. Can
> someone suggest a way to create such rank variable for subsets without
> using a loop?
>
> Thank you,
> Thomas
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>