plot.R

plot.R

p <- ggplot(data.frame(x=c(2,4,8,16), y=c(1, 1.84, 3.7, 7.22)), aes(x=x, y=y)) + 
  ggtitle("Scaling of Spark Decision Trees") +
  xlab("Machines") + 
  ylab("Speedup") + 
  
  geom_line(aes(color="actual")) + 
  geom_point() + 
  geom_text(aes(label=y), vjust=.5, hjust=-.15) +
  
  geom_line(aes(y=2*y, color="implied")) + 
  geom_point(aes(y=2*y)) + 
  geom_text(aes(y=2*y, label=2*y), vjust=1.4) + 
  
  geom_line(aes(x=a, y=b, color="ideal"), 
            data=data.frame(a = c(2,16), b = c(2,16))) + 
  geom_point(aes(x=a, y=b), 
             data=data.frame(a = c(4,8,16), b = c(4,8,16))) + 
  geom_text(aes(label=b, x=a, y=b), 
            data=data.frame(a = c(4,8,16), b = c(4,8,16)), vjust=-1) +
  
  geom_text(aes(label="actual", x=12, y=5.85, angle=24)) +
  geom_text(aes(label="implied", x=12.5, y=11.8, angle=41)) +
  geom_text(aes(label="ideal",
                x=12, y=12.5, angle=45)) +
  
  geom_segment(aes(x=15, y=10.6, xend=15, yend=13.25), arrow=arrow()) + 
  geom_text(aes(x=15, y=10.25, label="???")) +
  geom_segment(aes(x=15, y=9.9, xend=15, yend=7), arrow=arrow()) + 
  
  scale_x_continuous(limits=c(1,16), breaks=c(2,4,8,16)) +
  scale_y_continuous(limits=c(1,16), breaks=c(2,3.68,4,6,7.4,8,14.44,16), 
                     labels=c(2,"",4,6,"",8,"",16)) + 
  scale_colour_manual(name="", values=c("ideal"="darkgreen", "actual"="red", "implied"="blue")) +

  theme_fivethirtyeight() + coord_fixed()
  
gt <- ggplot_gtable(ggplot_build(p))
gt$layout$clip[gt$layout$name == "panel"] <- "off"
grid.draw(gt)