Drawing publish-quality scatterplots

Drawing a publication quality scatterplot doesn't require stacking up all that we've seen until now. It's usually the other way round. Telling a good history means sticking with the right tools and not deploying unnecessary ones. Unnecessary usually is synonymous to mixed signals. The history you need to tell with your plot may be a short or long one, may request few or many devices. This decision is up to you, but there are general things to look for that improves almost any scatterplot. 

All graphics brought until now by this chapter may be considered good results if those were made only for exploratory purposes. However, on the other hand, they can be considered unfinished work when it comes to publish quality standards-there is still a pretty run to make.

Jeff Leek stresses that defaults in ggplot2 are pretty enough that might trick you into thinking the graph is production ready by using only defaults. Each context will request a different amount of work to craft a publishing quality plot, but as Jeff Leek stands for, there at least a few things to look for that would improve almost any plot:

• Give complete information (measure included) in legends and labels

• Check spelling

• Grow labels bigger (increase font size)

• Grow axes bigger (include more breaks)

These are very basic, yet important points to visit when you seek expository quality. It is hardly reasonable to avoid any of them. The first thing to do in order to start is to brew a graphic using only defaults. Storing it into a variable and taking a good look at it is always a good way to begin.

This recipe's goal is to draw a nice exploratory graph that tells that different iris species hold different petal length and width relations, and turn it into a publication quality illustration. To the mission!