[R] Barchart in lattice - wrong order of groups, data labels on top of each other, and a legend question

[Deepayan Sarkar]

On Mon, May 18, 2009 at 11:47 AM, Dimitri Liakhovitski <ld7631 at gmail.com> wrote:
> Hello!
> I have a question about my lattice barchart that I am trying to build
> in Section 3 below. I can't figure out a couple of things:
> 1. When I look at the dataframe "test" that I am trying to plot, it
> looks right to me (the group "Total" is always the first out of 5).
> However, in the chart it is the last. Why?
> 2. How can I make sure the value labels (on y) are not sitting on top
> of each other but on top of the respective bar?
> 3. Is there any way to make the legend group items horizontally as
> opposed to now (vertically - taking up too much space)

For 1 and 3, use

         auto.key = list(points = FALSE,
                         rectangles = TRUE,
                         reverse.rows = TRUE,
                         columns = 2,
                         space = "bottom")

>From ?xyplot (under 'key'):

               'reverse.rows' logical, defaulting to 'FALSE'.  If
                    'TRUE', all components are reversed _after_ being
                    replicated (the details of which may depend on the
                    value of 'rep').  This is useful in certain
                    situations, e.g. with a grouped 'barchart' with
                    'stack = FALSE' with the categorical variable on
                    the vertical axis, where the bars in the plot will
                    usually be ordered from bottom to top, but the
                    corresponding legend will have the levels from top
                    to bottom (unless, of course, 'reverse.rows =
                    TRUE').  Note that in this case, unless all columns
                    have the same number or rows, they will no longer
                    be aligned.

               'columns' the number of columns column-blocks the key is
                    to be divided into, which are drawn side by side.


2 is hard with a simple custom panel function, because you need to
replicate some fairly involved calculations that are performed in
panel.barchart. Your best bet is to start with a copy of
panel.barchart, and then add calls to panel.text at suitable places.

-Deepayan


> Thanks a lot!
> Dimitri
>
> ### Section 1: generates my data set "data" - just run: #####
>
> N<-100
> myset1<-c(1,2,3,4,5)
> probs1<-c(.05,.10,.15,.40,.30)
> myset2<-c(0,1)
> probs2<-c(.65,.30)
> myset3<-c(1,2,3,4,5,6,7)
> probs3<-c(.02,.03,.10,.15,.20,.30,.20)
>
> group<-unlist(lapply(1:4,function(x){
>        out<-rep(x,25)
>        return(out)
> }))
> set.seed(1)
> a<-sample(myset1, N, replace = TRUE,probs1)
> a[which(rbinom(100,2,.01)==1)]<-NA
> set.seed(12)
> b<-sample(myset1, N, replace = TRUE,probs1)
> b[which(rbinom(100,2,.01)==1)]<-NA
> set.seed(123)
> c<-sample(myset2, N, replace = TRUE,probs2)
> set.seed(1234)
> d<-sample(myset2, N, replace = TRUE,probs2)
> set.seed(12345)
> e<-sample(myset3, N, replace = TRUE,probs3)
> e[which(rbinom(100,2,.01)==1)]<-NA
> set.seed(123456)
> f<-sample(myset3, N, replace = TRUE,probs3)
> f[which(rbinom(100,2,.01)==1)]<-NA
> data<-data.frame(group,a=a,b=b,c=c,d=d,e=e,f=f)
> data["group"]<-lapply(data["group"],function(x) {
>        x[x %in% 1]<-"Group 1"
>        x[x %in% 2]<-"Group 2"
>        x[x %in% 3]<-"Group 3"
>        x[x %in% 4]<-"Group 4"
>        return(x)
> })
> data$group<-as.factor(data$group)
> lapply(data,table,exclude=NULL)
>
> tables<-lapply(data,function(x){
>        out<-table(x)
>        out<-prop.table(out)
>        out<-round(out,3)*100
>        return(out)
> })
> str(tables[2])
>
> ##### Section 2: Generating a list of tables with percentages to be
> plotted in barcharts - just run: #####
>
> listoftables<-list()
> for(i in 1:(length(data)-1)) {
>  listoftables[[i]]<-data.frame()
> }
> for(i in 1:length(listoftables)) {
>    total<-table(data[[i+1]])
>    groups<-table(data[[1]],data[[i+1]])
>    total.percents<-as.data.frame(t(as.vector(round(total*100/sum(total),1))))
>    groups.percents<-as.data.frame(t(apply(groups,1,function(x){
>      out<-round(x*100/sum(x),1)
>     return(out)
>  })))
>  names(total.percents)<-names(groups.percents)
>  final.table<-rbind(total.percents,groups.percents)
>  row.names(final.table)[1]<-"Total"
>  final.table<-as.matrix(final.table)
>  listoftables[[i]]<-final.table
> }
> names(listoftables)<-names(data)[2:(length(listoftables)+1)]
>
>
> ### Section 3 - building the graph for the very first table of the
> "listoftables" ###
> library(lattice)
> i<-1
> test <- data.frame(Group = rep(row.names(listoftables[[i]]),5), a =
> rep(1:5,each=5),Percentage = as.vector(listoftables[[i]]))
> par.settings=trellis.par.set(reference.line = list(col = "gray", lty ="dotted"))
> barchart(Percentage~a, test, groups = Group, horizontal = F,
> auto.key = list(points = FALSE, rectangles = TRUE, space =
> "bottom"),ylim = c(0,50),
>    panel = function(y,x,...) {
>    panel.grid(h = -1, v = -1)
>    panel.barchart(x, y, ...)
>    ltext(x, y, labels=round(y,0),cex=.7,col="black",font=2,pos=3)
> })
>
>
> --
> Dimitri Liakhovitski
> MarketTools, Inc.
> Dimitri.Liakhovitski at markettools.com
>
> ______________________________________________
> 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.
>