Three factor plotting using xyplot

I had a problem with ggplot that I am not able to solve, so maybe someone here can point out the reason. Sorry that I am not able to upload my dataset, but some data description can be found below. The output of the ggplot is shown below, except NO line, every other thing is OK.

> all.data<-read.table("D:/PAM/data/Rural_Recovery_Edit.csv",head=T,sep=",")
> all.data$Water<-factor(all.data$Water,labels=c("W30","W60","W90"))
> all.data$Polymer<-factor(all.data$Polymer,labels=c("PAM-0  ","PAM-10  ","PAM-40  "))
> all.data$Group<-factor(all.data$Group,labels=c("Day20","Day25","Day30"))
> dat<-data.frame(Waterconsump=all.data[,9],Water=all.data$Water,Polymer=all.data$Polymer,Age=all.data$Group)

> ggplot(dat,aes(x=Water,y=Waterconsump,colour=Polymer))+
+ stat_summary(fun.y=mean, geom="line",size=2)+
+ stat_summary(fun.ymin=min,fun.ymax=max,geom="errorbar")+#,position="dodge"
+ facet_grid(~Age)

> dim(dat)
[1] 108   4
> head(dat)
  Waterconsump Water  Polymer   Age
1         10.5   W30 PAM-10   Day20
2         10.3   W30 PAM-10   Day20
3         10.1   W3开发者_JS百科0 PAM-10   Day20
4          7.7   W30 PAM-10   Day20
5          8.6   W60 PAM-10   Day20
6          8.4   W60 PAM-10   Day20
> table(dat$Water)

W30 W60 W90 
 36  36  36 
> table(dat$Polymer)

 PAM-0   PAM-10   PAM-40   
      36       36       36 
> table(dat$Age)

Day20 Day25 Day30 
   36    36    36 

below is the background for this Q

I would like to plot several variables that were subjected to the same, 3 factors. Using xyplot, I am able to plot 2 of them, within one figure. However, I have no idea how to include the third, and arrange the figure into N subplots (N equals the level number of the third factor). So, my aims would be:

1. Plot the 3rd facotors, and split the plot into N subplots, where N is the levels of the 3rd factor.

2. Better to work as a function, as I need to plot a several variables. Below is the example figure with only two factors, and my working example to plot 2 factors.

Thanks in advance~

Marco

library(reshape)
library(agricolae)
library(lattice)
yr<-gl(10,3,90:99)
trt<-gl(4,75,labels=c("A","B","C","D"))

third<-gl(3,100,lables=c("T","P","Q")) ### The third factor to split the figure in to 4 subplots

dat<-cbind(runif(300),runif(300,min=1,max=10),runif(300,min=100,max=200),runif(300,min=1000,max=1500))
colnames(dat)<-paste("Item",1:4,sep="-")
fac<-factor(paste(trt,yr,sep="-"))
dataov<-aov(dat[,1]~fac)
dathsd<-sort_df(HSD.test(dataov,'fac'),'trt')
trtplt<-gl(3,10,30,labels=c("A","B","C"))
yrplt<-factor(substr(dathsd$trt,3,4))

prepanel.ci <- function(x, y, ly, uy, subscripts, ...) 
{ 
    x <- as.numeric(x) 
    ly <- as.numeric(ly[subscripts]) 
    uy <- as.numeric(uy[subscripts]) 
    list(ylim = range(y, uy, ly, finite = TRUE)) 
} 
panel.ci <- function(x, y, ly, uy, subscripts, pch = 16, ...) 
{ 
    x <- as.numeric(x) 
    y <- as.numeric(y) 
    ly <- as.numeric(ly[subscripts]) 
    uy <- as.numeric(uy[subscripts]) 
    panel.arrows(x, ly, x, uy, col = "black", 
                 length = 0.25, unit = "native", 
                 angle = 90, code = 3) 
    panel.xyplot(x, y, pch = pch, ...) 
} 

xyplot(dathsd$means~yrplt,group=trtplt,type=list("l","p"),
        ly=dathsd$means-dathsd$std.err,
        uy=dathsd$means+dathsd$std.err,
        prepanel = prepanel.ci, 
        panel = panel.superpose, 
        panel.groups = panel.ci 
        )

Here is another way of doing it, using the magic of ggplot. Because ggplot will calculate summaries for you, I suspect it means you can skip the entire step of doing aov.

The key is that your data should be in single data.frame that you can pass to ggplot. Note that I have created new sample data to demonstrate.

library(ggplot2)

df <- data.frame(
  value = runif(300),
  yr = rep(1:10, each=3),
  trt = rep(LETTERS[1:4], each=75),
  third = rep(c("T", "P", "Q"), each=100)
)

ggplot(df, aes(x=yr, y=value, colour=trt)) + 
  stat_summary(fun.y=mean, geom="line", size=2) +
  stat_summary(fun.ymin=min, fun.ymax=max, geom="errorbar") +
  facet_grid(~third)

You can go one step further and produce facets in two dimensions:

ggplot(df, aes(x=yr, y=value, colour=trt)) + 
  stat_summary(fun.y=mean, geom="line", size=2) +
  stat_summary(fun.ymin=min, fun.ymax=max, geom="errorbar") +
  facet_grid(trt~third)

This gets pretty close, but I forget how to colour the error lines using the group variable in Lattice and Deepayan's book is at work.

## format a new data structure with all variables we want
dat <- data.frame(dathsd[, c(2,5)], treat = trtplt, yrplt = yrplt,
                  upr = dathsd$means + 2 * dathsd$std.err,
                  lwr = dathsd$means - 2 * dathsd$std.err)
## compute ylims
ylims <- range(dat$lwr, dat$upr)
ylims <- ylims + (c(-1,1) * (0.05 * diff(ylims)))
## plot
xyplot(means ~ yrplt, data = dat, group = treat, lwr = dat$lwr, upr = dat$upr,
       type = c("p","l"), ylim = ylims,
       panel = function(x, y, lwr, upr, ...) {
           panel.arrows(x0 = x, y0 = lwr, x1 = x, y1 = upr,
                        angle = 90, code = 3, length = 0.05)
           panel.xyplot(x, y, ...)
       })

And produces: