I have been playing with bi-variate choropleth maps and have gotten stuck on how to create a 2d legend similar to the one by Joshua Stevens shown here:
To illus
#test desired legend appearance
library(ggplot2)
library(reshape2)
#use color scheme shown here http://www.joshuastevens.net/cartography/make-a-bivariate-choropleth-map/
bvColors=c("#be64ac","#8c62aa","#3b4994","#dfb0d6","#a5add3","#5698b9","#e8e8e8","#ace4e4","#5ac8c8")
melt(matrix(1:9,nrow=3))
legendGoal=melt(matrix(1:9,nrow=3))
test<-ggplot(legendGoal, aes(Var2,Var1,fill = as.factor(value)))+ geom_tile()
test<- test + scale_fill_manual(name="More Var2 -->",values=bvColors,drop=FALSE)
test<-test+guides(fill = guide_legend(nrow = 3))
test<-test + theme(legend.text=element_blank())
test
The only remaining trick is to find a way to add some vertical on the side of the legend saying "More Var1 -->." Here's a butt-ugly way to do it:
test<-ggplot(legendGoal, aes(Var2,Var1,fill = as.factor(value)))+ geom_tile()
test<- test + scale_fill_manual(name="More Var2 -->",values=bvColors,labels=c("","","","","","","More","Var 1"," v "))
test<-test+guides(fill = guide_legend(nrow = 3))
#test<-test + theme(legend.text=element_blank())
test
But, as zx shows, extending ggplot2 with the cowplot package is the complete solution:
#test desired legend appearance
library(ggplot2)
library(cowplot)
library(reshape2)
#use color scheme shown here http://www.joshuastevens.net/cartography/make-a-bivariate-choropleth-map/
bvColors=c("#be64ac","#8c62aa","#3b4994","#dfb0d6","#a5add3","#5698b9","#e8e8e8","#ace4e4","#5ac8c8")
melt(matrix(1:9,nrow=3))
legendGoal=melt(matrix(1:9,nrow=3))
test<-ggplot(legendGoal, aes(Var2,Var1,fill = as.factor(value)))+ geom_tile()
test<- test + scale_fill_manual(name="",values=bvColors)
test<-test+guides(fill = guide_legend(nrow = 3))
test<-test + theme(legend.text=element_blank())
test<-ggdraw(test) + draw_text(text = "More Var 2 -->",x=0.91,y=0.58)
test<-ggdraw(test) + draw_text(text = "More Var 1 -->",x=0.84,y=0.5,angle=270)
test
Just for fun this is the map that I made with this technique: