sizetree: Display a hierarchical breakdown of disjunct categories

Description

Display a data frame in which the values in each successive column represent subcategories of the previous column as stacked rectangles.

Usage

sizetree(x,left=0,top,right=1,lastcenter=NA,showval=TRUE,showcount=TRUE,
  firstcall=TRUE,col=NA,colorindex=1,...)

Arguments

A data frame in which each successive column represents subcategories of the previous column.

left

The left edge of the current stack of rectangles in user units.

top

The top of the current stack of rectangles in user units.

right

The right edge of the current stack of rectangles in user units.

lastcenter

The center of the previous rectangle from which the next breakdown of categories arises. There is almost no reason to change it.

showval

Whether to display the values representing the categories.

showcount

Whether to display the count for the categories.

firstcall

A flag for the function - do not alter this.

col

Optional fill colors for the rectangles. See Details

colorindex

The index for col if it is a list of color vectors.

...

additional arguments passed to plot.

Value

Details

sizetree displays disjunct hierarchical categories as stacked rectangles. It accepts a data frame in which the values in the first column represent categories, the values in the second column represent subcategories of the first column, and so on. The first column will be displayed as a stack of rectangles, the height of each proportional to the count for each category. Each substack of rectangles in the second stack will represent the breakdown of counts for its superordinate category and so on through the columns. Empty categories are ignored and NAs will produce gaps, which will propagate across subsequent stacks. Typically, the user will simply pass the data frame, which should only contain columns that are hierarchical categories, set showval and showcount to the desired values, and pass colors for the top level categories if these are wanted. If colors are passed, it is best to have at least as many colors as there are categories in the first column or some will be recycled. If different colors are desired for different levels, as in the example, pass a list of colors. If colors are to be matched across columns in which the number of levels is different (e.g. some levels are missing in the sub-categories), the color list must reflect this and the levels of the factors must be consistent with the colors - see the example. The firstcall argument is necessary for the function to initialize the plot, as each breakdown involves a recursive call. If it is changed, the best that can be expected is an uninformative plot.

Examples

Run this code

cat1<-factor(sample(c("None","Low","Medium","High","Extreme"),40,TRUE),
  levels=c("None","Low","Medium","High","Extreme"))
 cat2<-factor(sample(c("None","Low","Medium","High"),40,TRUE),
  levels=c("None","Low","Medium","High"))
 cat3<-factor(sample(c("None","Low","High"),40,TRUE),
  levels=c("None","Low","High"))
 hcats<-data.frame(cat1,cat2,cat3)
 # throw in a few NAs
 hcats$cat1[10]<-NA
 hcats$cat2[c(15,20)]<-NA
 hcats$cat3[c(11,14,25)]<-NA
 bhcol<-list(c("#ff8080","#dddd80","#80ff80","#0000ff","#80dddd"),
  c("#ff8080","#dddd80","#80ff80","#0000ff"),
  c("#ff8080","#dddd80","#0000ff"))
 sizetree(hcats,col=bhcol,main="Hierarchical count chart (sizetree)")