R help - Sep 2007 - Getting group-wise standard scores of a vector

Hi,

I want to be able to create a vector of z-scores from a vector of 
continuous data, conditional on a group membership vector.

Say you have 20 numbers distributed normally with a mean of 50 and an sd 
of 10:

x <- rnorm(20, 50, 10)


Then you have a vector that delineates 2 groups within x:

group <- sort(rep(c("A", "B"), 10))

test.data <- data.frame(cbind(x, group))

I know that if you break up the x vector into 2 different vectors then 
it becomes easy to calculate the z scores for each vector, then you 
stack them and append them to the original
data frame.  Is there anyway to apply this sort of calculation without 
splitting the original vector up?  I tried a really complex ifelse 
statement but it didn't seem to work.

Thanks in advance,
Matthew Dubins

Hello -

First, I doubt you really want to cbind() those two vectors within the 
data.frame() function call.

test.data <- data.frame(x, group)  is probably what you want.  That may 
be the source of your trouble.

If you really want a vector returned, the following should work given 
your test.data is constructed without the cbind():

unlist(by(test.data$x, test.data$group, function(x) (x - mean(x)) / 
sd(x)), use.names = FALSE)

Is that what you're after?

Erik


Matthew Dubins wrote:
> Hi,
> 
> I want to be able to create a vector of z-scores from a vector of 
> continuous data, conditional on a group membership vector.
> 
> Say you have 20 numbers distributed normally with a mean of 50 and an sd 
> of 10:
> 
> x <- rnorm(20, 50, 10)
> 
> 
> Then you have a vector that delineates 2 groups within x:
> 
> group <- sort(rep(c("A", "B"), 10))
> 
> test.data <- data.frame(cbind(x, group))
> 
> I know that if you break up the x vector into 2 different vectors then 
> it becomes easy to calculate the z scores for each vector, then you 
> stack them and append them to the original
> data frame.  Is there anyway to apply this sort of calculation without 
> splitting the original vector up?  I tried a really complex ifelse 
> statement but it didn't seem to work.
> 
> Thanks in advance,
> Matthew Dubins
> 
> ______________________________________________
> 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.

tapply is also very useful: 


my.df<-data.frame(x=rnorm(20, 50, 10),group=factor(sort(rep(c("A",
"B"), 10))))
tapply(my.df$x,my.df$group,function(x){(x-mean(x))/sd(x)})



-----Original Message-----
From: r-help-bounces at r-project.org
[mailto:r-help-bounces at r-project.org]On Behalf Of Matthew Dubins
Sent: 26 September 2007 21:57
To: r-help at r-project.org
Subject: [R] Getting group-wise standard scores of a vector


Hi,

I want to be able to create a vector of z-scores from a vector of 
continuous data, conditional on a group membership vector.

Say you have 20 numbers distributed normally with a mean of 50 and an sd 
of 10:

x <- rnorm(20, 50, 10)


Then you have a vector that delineates 2 groups within x:

group <- sort(rep(c("A", "B"), 10))

test.data <- data.frame(cbind(x, group))

I know that if you break up the x vector into 2 different vectors then 
it becomes easy to calculate the z scores for each vector, then you 
stack them and append them to the original
data frame.  Is there anyway to apply this sort of calculation without 
splitting the original vector up?  I tried a really complex ifelse 
statement but it didn't seem to work.

Thanks in advance,
Matthew Dubins

______________________________________________
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.