Grammar of Graphics
Posted by Jordan Kersey
You must use good grammer!
Generally speaking, the grammer of graphics implies the language and cues used to create meaningful, and descriptive graphic representations of the raw data. “Bad grammer”, as it does in the real-world, leads to inaccurate portrayal of the idea, and opens the door to potential confusion. However, with the use of “good grammer”, one can expertly express what they are trying to, and is likely to allow others to interpret a great understanding of the material.
Due to the structure of good graphic grammer, it enables one to manipulate and design very extensive and creative plots. By having seven different classes of descriptive method, one can produce a graphic tailored to their data needs. However, this does not come without its challenges.
One challenge is that the system must be applied in a specific order. By augmenting the desired graphic in the wrong order (addressing a class before it is called for), one will produce a statistically meaningless graph. Which, essentially ruins the analysis of the data set.
A second challenge could be how to correctly apply the data set one is working with, to the good grammer method described. Each data set is unique, and therefore, must be considered in its own way. Familiarity with ones data, and having tidy data is benefical in this instance.
We have used grammer of graphics methods during class labs, when we have taken random data sets, and attempted to produce a graph which makes sense, and clearly depicts what we are trying to express. Also, in homework 3, it was necessary to tweak the “standard” plots, in order to have the correct relationships displayed. During labs, we have worked with adding additional code, to hone-in on a specific piece which was muddled by the simple, original graph.
Due to it’s complexity, the grammer of graphics is something that will take time to fully grasp, and be able to use at one’s disposal. However, if mastered, the grammer of graphics (using such packages as ggplot2), has the potential for any scientist to become an artist at depicting their data!