---------- Forwarded message ----------
From: <r-help-bounces@r-project.org>
Date: Mon, May 11, 2009 at 10:24 PM
Subject: The results of your email commands
To: das.moumita.online@gmail.com
The results of your email command are provided below. Attached is your
original message.
- Results:
Ignoring non-text/plain MIME parts
- Unprocessed:
What is the function for partial correlation.The function which i found
here:-- http://www.yilab.gatech.edu/pcor.R and used the recursive
version, showed errors while computing partial correlation,where the
third
variable i.e controlled variable for a pair is more than that .That is
when
i had more than one third variable ,it always became an infinite loop,it
else.
# By using recursive formula
pcor.rec <- function(x,y,z,method="p",na.rm=T){
#
x <- c(x)
y <- c(y)
z <- as.data.frame(z)
if(dim(z)[2] == 0){
stop("There should be given data\n")
}
data <- data.frame(x,y,z)
if(na.rm == T){
data = na.omit(data)
}
- Ignored:
# recursive formula
if(dim(z)[2] == 1){
tdata <- na.omit(data.frame(data[,1],data[,2]))
rxy <- cor(tdata[,1],tdata[,2],m=method)
tdata <- na.omit(data.frame(data[,1],data[,-c(1,2)]))
rxz <- cor(tdata[,1],tdata[,2],m=method)
tdata <- na.omit(data.frame(data[,2],data[,-c(1,2)]))
ryz <- cor(tdata[,1],tdata[,2],m=method)
rxy.z <- (rxy - rxz*ryz)/( sqrt(1-rxz^2)*sqrt(1-ryz^2) )
return(rxy.z)
}else{
x <- c(data[,1])
y <- c(data[,2])
z0 <- c(data[,3])
zc <- as.data.frame(data[,-c(1,2,3)])
rxy.zc <- pcor.rec(x,y,zc,method=method,na.rm=na.rm)
rxz0.zc <- pcor.rec(x,z0,zc,method=method,na.rm=na.rm)
ryz0.zc <- pcor.rec(y,z0,zc,method=method,na.rm=na.rm)
rxy.z <- (rxy.zc - rxz0.zc*ryz0.zc)/(
sqrt(1-rxz0.zc^2)*sqrt(1-ryz0.zc^2) )
return(rxy.z)
}
}
and this piece of code also i tried to use.
Rinv <- solve
<http://wiki.r-project.org/rwiki/rhelp.php?id=solve>(R)
D <- diag <http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(1 /
sqrt
<http://wiki.r-project.org/rwiki/rhelp.php?id=sqrt>(diag
<http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(Rinv)))
P <- -D <http://wiki.r-project.org/rwiki/rhelp.php?id=D> %*% Rinv
%*%
D <http://wiki.r-project.org/rwiki/rhelp.php?id=D>
where R is teh correlation matrix.
but NaN error at for
sqrt <http://wiki.r-project.org/rwiki/rhelp.php?id=sqrt>(diag
<http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(Rinv) were
generated.
http://wiki.r-project.org/rwiki/doku.php?id=tips:data-matrices:part_corr&s=correlation
Any convinient function for partial correlation for both one
controlled variable or more.
--
Thanks
Moumita
- Done.
---------- Forwarded message ----------
From: Moumita Das <das.moumita.online@gmail.com>
To: r-help-request@r-project.org
Date: Mon, 11 May 2009 22:23:33 +0530
Subject: partial correlation function
Hi friends,
What is the function for partial correlation.The function which i found
here:-- http://www.yilab.gatech.edu/pcor.R and used the recursive
version, showed errors while computing partial correlation,where the third
variable i.e controlled variable for a pair is more than that .That is when
i had more than one third variable ,it always became an infinite loop,it
else.
# By using recursive formula
pcor.rec <- function(x,y,z,method="p",na.rm=T){
#
x <- c(x)
y <- c(y)
z <- as.data.frame(z)
if(dim(z)[2] == 0){
stop("There should be given data\n")
}
data <- data.frame(x,y,z)
if(na.rm == T){
data = na.omit(data)
}
# recursive formula
if(dim(z)[2] == 1){
tdata <- na.omit(data.frame(data[,1],data[,2]))
rxy <- cor(tdata[,1],tdata[,2],m=method)
tdata <- na.omit(data.frame(data[,1],data[,-c(1,2)]))
rxz <- cor(tdata[,1],tdata[,2],m=method)
tdata <- na.omit(data.frame(data[,2],data[,-c(1,2)]))
ryz <- cor(tdata[,1],tdata[,2],m=method)
rxy.z <- (rxy - rxz*ryz)/( sqrt(1-rxz^2)*sqrt(1-ryz^2) )
return(rxy.z)
}else{
x <- c(data[,1])
y <- c(data[,2])
z0 <- c(data[,3])
zc <- as.data.frame(data[,-c(1,2,3)])
rxy.zc <- pcor.rec(x,y,zc,method=method,na.rm=na.rm)
rxz0.zc <- pcor.rec(x,z0,zc,method=method,na.rm=na.rm)
ryz0.zc <- pcor.rec(y,z0,zc,method=method,na.rm=na.rm)
rxy.z <- (rxy.zc - rxz0.zc*ryz0.zc)/( sqrt(1-rxz0.zc^2)*sqrt(1-ryz0.zc^2) )
return(rxy.z)
}
}
and this piece of code also i tried to use.
Rinv <- solve <http://wiki.r-project.org/rwiki/rhelp.php?id=solve>(R)
D <- diag <http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(1 / sqrt
<http://wiki.r-project.org/rwiki/rhelp.php?id=sqrt>(diag
<http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(Rinv)))
P <- -D <http://wiki.r-project.org/rwiki/rhelp.php?id=D> %*% Rinv %*%
D <http://wiki.r-project.org/rwiki/rhelp.php?id=D>
where R is teh correlation matrix.
but NaN error at for
sqrt <http://wiki.r-project.org/rwiki/rhelp.php?id=sqrt>(diag
<http://wiki.r-project.org/rwiki/rhelp.php?id=diag>(Rinv) were
generated.
http://wiki.r-project.org/rwiki/doku.php?id=tips:data-matrices:part_corr&s=correlation
Any convinient function for partial correlation for both one
controlled variable or more.
--
Thanks
Moumita
--
Thanks
Moumita
[[alternative HTML version deleted]]
