by Justin Gosses
This visualization is based heavily on Jason Davies d3-parasets project, which in turn was based on based on Robert Kosara windows and MacOS X native GUI. You can find a longer discussion on add-ons I want to build into this project in this blog post. One improvement I added in this d3.js implementation was the ability of the user to select which data categories or dimensions are shown. Additionally, the user can limit the visualizaion to only battles where a specific value in a specific category is present. For example, only the battles where the defender was the House Stark.
The rational for these features is that parallel set visualizations can quickly get messy and being able to quickly explore different combinations of data without any coding or data preparation is makes this type of data visualization more useful.
This was devised as a data exploration tool. It helps to think of specific questions when using it. One example might be, "do attackers tend to be more successful in siege versus pitched battle?" or "How do the Starks fare in ambushes in year 300?" If you are limiting a category to only one or two values using the secondary drop-down menu, move those categories to the bottom of the visualization to see data relationships easier. Arrange the rest of the data categories vertically in a pseudo sentence structure.
The dataset for this visualization comes from Kaggle, a data science website geared to competition and community. Please check the original data source to see if there are any updates as the Game of Thrones story progresses. You can also download the lightly edited version used in this page here.
You can download the code for this data visualization from this github repo.
or read more about it in this blog post.