Dotplot with boxplot group summaries

Question

I'm plotting data by two categories, first the underlying data:

library(plyr)
library(ggplot2)

d1 <- data.frame(v1 = rnorm(30),
                  x = rep(letters[1:6],  
                          c(6, 3, 8, 3, 4, 6)),
              group = rep(paste("g", 1:3, sep = ""),
                          c(9, 11, 10)))

Then I order factors

of <- ddply(d1, .(x), function(i) mean(i$v1))
of <- of[order(-of$V1),]
d1$x <- factor(d1$x, levels = of$x)

And this provides the following plot

ggplot(d1) + geom_point(aes( x = v1, y = x)) + 
             facet_grid(group ~ .,
                        scales = "free_y",
                        space = "free_y")

But thne I also want to add a separate horizontal boxplot for each group, depicting the relevant distribution. I know the first step is to provide a "Total" space in the y axis:

d2 <- rbind(d1, 
        data.frame(v1 = NA,
                    x = "Total",
                group = unique(d1$group)))

 ggplot(d2) + geom_point(aes( x = v1, y = x)) + 
              facet_grid(group ~ .,
                         scales = "free_y",
                         space = "free_y")

Which provides the following, and here I'm stuck

I just want a horizontal boxplot next to the "Total" axis break, summarizing each group's data.

Thanks in advance.

Answer 1

Starting point would be to make two data frame - d1 as your original data frame and d2 where x values all are Total.

d2 <- d1
d2$x <- "Total"

Then use d1 to make points and d2 to make boxplots. Here is example with vertical alignment of boxplots where x values are used as x. Problem is that to have horizontal boxplot you have to use coord_flip(). Flipped axis and free scales can't be used together.

ggplot()+geom_point(data=d1,aes(x=x,y=v1))+
  geom_boxplot(data=d2,aes(x=x,y=v1))+
  facet_wrap(~group ,
             scales = "free_x")

Workaround to get in direction you need would be to make three separate plots (one for each group) and then put them together with grid.arrange() for library gridExtra. For the first and second plot I removed all ticks, texts and titles for x axis.

library(gridExtra)

p1<-ggplot()+geom_point(data=subset(d1,group=="g1"),aes(x=x,y=v1))+
  geom_boxplot(data=subset(d2,group=="g1"),aes(x=x,y=v1))+
  coord_flip()+
  facet_grid(group~.)+
  scale_y_continuous(limits=c(-2,2.5))+
  theme(axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank())

p2<-ggplot()+geom_point(data=subset(d1,group=="g2"),aes(x=x,y=v1))+
  geom_boxplot(data=subset(d2,group=="g2"),aes(x=x,y=v1))+
  coord_flip()+
  facet_grid(group~.)+
  scale_y_continuous(limits=c(-2,2.5))+
  theme(axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank())

p3<-ggplot()+geom_point(data=subset(d1,group=="g3"),aes(x=x,y=v1))+
  geom_boxplot(data=subset(d2,group=="g3"),aes(x=x,y=v1))+
  coord_flip()+
  facet_grid(group~.)+
  scale_y_continuous(limits=c(-2,2.5))

grid.arrange(p1,p2,p3)