Dear All,
Sorry to bother you again. As my previous mail was messy to understand,
please find it again to give me a solution. I'd like to do a partial
correaltion test ['pcor.test ()' or 'parcor()'] between Irid.area and
Casa.PC1 variables controlling the influence of SL (co-variate) according
to categorical group factors (Quantity or Quality group).
Can anyone give me any example or source of package that will be
appreciated. I already tried with "ggm" (doesn't work in R now), "ppcor",
but failed to modify or adjust with my data to anlyse as I want.
My effort:
> data1 Quantity Quality SL Irid.area Casa.PC1
1 High Low 16.38 10.31 1.71173956
2 High High 15.95 16.52 0.01338354
3 High High 15.69 12.74 2.22849088
4 High Low 14.76 9.80 1.55497583
5 High Low 14.63 12.95 1.82376797
6 High High 14.32 14.21 3.15205984
7 High High 14.95 12.57 2.06926504
8 High Low 15.37 13.55 1.88602742
9 High Low 14.73 14.18 1.12744060
10 High High 16.08 15.98 1.43556331
11 Low Low 13.95 16.05 -1.44961267
12 Low Low 14.03 12.58 -1.68596884
13 Low High 14.82 13.57 -0.09742642
14 Low High 14.32 12.16 -1.40351201
15 Low Low 14.33 7.66 -1.33665471
16 Low Low 15.01 10.15 -1.25701927
17 Low High 14.01 9.79 -0.71540450
18 Low Low 14.25 17.38 -1.29695402
19 Low High 14.55 16.11 -0.61689594
20 Low High 13.98 11.49 -0.65401736
> ### Correlation test according to group> library("MASS")> with(data1,
> cor.test(~ Irid.area + Casa.PC1, subset=(Quantity=="High")))# gives cor,
> df+2, p-values
Pearson's product-moment correlation
data: Irid.area and Casa.PC1
t = -1.0507, df = 8, p-value = 0.3241
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
-0.8020154 0.3604104
sample estimates:
cor
-0.3482398
> with(data1, cor.test(~ Irid.area + Casa.PC1, subset=(Quantity=="Low")))#
> gives cor, df+2, p-values
Pearson's product-moment correlation
data: Irid.area and Casa.PC1
t = 0.1209, df = 8, p-value = 0.9068
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
-0.6031431 0.6547228
sample estimates:
cor
0.04269795
> ### ppcor tests:> y.data <- data.frame(+ v1=data1$Irid.area,+
> v2=data1$Casa.PC1,+ f1=data1$Quantity,+ f2=data1$Quality)>
> library(ppcor)> pcor(y.data)Error in pcor(y.data) : 'x' must be numeric ## I
> also tried it with numeric values, but didn't give me the results same as
> JMP> # partial correlation between "v1" and "v2" given "f1" and "f2">
> pcor.test(y.data$v1,y.data$v2,y.data[,c("f1","f2")])Error: is.numeric(y) ||
> is.logical(y) is not TRUE???
Then I tried with
> #####pcor with ggm#####> library(ggm)# suggested by Andy Field et al. on
> their book "Discovering statistics using R".Loading required package:
> graphError: package graph could not be loaded> examData2<- data1[,
> c("Irid.area", "Casa.PC1", "SL")]> maleExam<-subset(data1, Quantity ==
> "High", select= c("Irid.area", "Casa.PC1"))> femaleExam<-subset(data1,
> Quantity == "Low", select= c("Irid.area", "Casa.PC1"))> cor(maleExam)
> Irid.area Casa.PC1
Irid.area 1.0000000 -0.3482398
Casa.PC1 -0.3482398 1.0000000> cor(femaleExam) Irid.area Casa.PC1
Irid.area 1.00000000 0.04269795
Casa.PC1 0.04269795 1.00000000> # partial correlation between two
var.> pcor.test(examData2$Irid.area, examData2$Casa.PC1,
examData2$SL..COVARIATE.,Quantity == "High")Error in
pcor.test(examData2$Irid.area, examData2$Casa.PC1,
examData2$SL..COVARIATE., :
'use' should be either "rec" or "mat"!In addition: Warning
messages:1: In if (use == "mat") { :
the condition has length > 1 and only the first element will be
used2: In if (use == "rec") { :
the condition has length > 1 and only the first element will be used
So, again can anyone help me to find out the solution, and sorry in advance
to disturb you with the same issue.
Cheers,
Jewel
e-mail:moshiu...@yahoo.com
g-mail: mrahmankuf...@gmail.com
[[alternative HTML version deleted]]
______________________________________________
R-help@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.
