Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

ggsave losing unicode characters from ggplot+gridExtra

Tags:

r

unicode

ggplot2

More code than you really need, but to set the mood:

#Make some data and load packages
data<-data.frame(pchange=runif(80,0,1),group=factor(sample(c(1,2,3),80,replace=T)))
library(dplyr)
library(magrittr)
library(gridExtra)
library(ggplot2)
data%<>%arrange(group,pchange) %>% mutate(num=1:80)

#Make plot that includes unicode characters
g1<-ggplot(data, aes(factor(num),pchange, fill = group,width=.4)) +
  geom_bar(stat="identity", position = "dodge") +
  theme_classic()+
  theme(axis.ticks = element_blank(), 
        axis.text.x = element_blank(),
        legend.position="right")+
  scale_y_continuous(breaks=c(0,.25,.5,.75,1))+
  xlab("")+
  scale_fill_discrete("Arbitrary Group",
                      breaks=c(1,2,3),
                      labels=c("< 1 Year", "\u2265 1 Year & \n\u2264 5 Years","> 5 Years"))


#I want to add an A below the plot (this may not be necessary for the issue, but its a part of the workflow so I thought I'd include it.
g <- arrangeGrob(plot=g1,
                 sub = textGrob("A",
                                x = .1,
                                hjust = .5,
                                vjust=-2,
                                gp = gpar(fontface = "bold",
                                          fontsize = 16,
                                          col="black")))

#Save the plot
ggsave(filename="X:/yourpath/Plot1.pdf", plot=g,
       width = 8, height = 4, units = "in", dpi = 600)

Here's what it looks like: Actual Plot

Here's what it should look like (in terms of the characters in the key; plot taken as jpeg directly from Rstudio plot window): Ideal Plot

like image 930
Andrew Taylor Avatar asked Feb 26 '15 15:02

Andrew Taylor


1 Answers

You have two options. One, use the cairo_pdf device instead of the default pdf in your call you ggsave, e.g.,

library(Cairo)
ggsave(filename="X:/yourpath/Plot1.pdf", plot=g, device=cairo_pdf,
       width = 8, height = 4, units = "in", dpi = 600)

The other option would be to use expressions instead of explicit unicode characters:

g<-ggplot(data, aes(factor(num),pchange, fill = group,width=.4)) +
  geom_bar(stat="identity", position = "dodge") +
  theme_classic()+
  theme(axis.ticks = element_blank(), 
        axis.text.x = element_blank(),
        legend.position="right")+
  scale_y_continuous(breaks=c(0,.25,.5,.75,1))+
  xlab("")+
  scale_fill_discrete("Arbitrary Group",
                      breaks=c(1,2,3),
                      labels=c(expression(phantom(0) < "1 Year"), 
                              expression(paste(phantom(0) >= "1 Year &", phantom(0) <= "5 Years")),
                              expression(phantom(0) > "5 Years")))



ggsave(filename="Plot1.pdf", plot=g,
       width = 8, height = 4, units = "in", dpi = 600)

Although, as you can see, with the second option the formatting isn't as tight as you might like.

As to why you are experiencing this issue, according to the answer here, the pdf driver can only handle single byte encodings.

enter image description here

like image 82
Matthew Plourde Avatar answered Oct 05 '22 12:10

Matthew Plourde