It’s not bad, but the different unit types appear as embedded subheaders, and a few variables contain compound values. The data on the site looks like this: click to enlarge

Adding text and arrow annotations to the plot to make the interpretation more explicit.

Text labels for points of the scatterplot with top and bottom values.

A somewhat hacky replacement of bars with crossbars for showing the contribution and correlations of the different variables to the PCA.

Leveraging gghighlight to show the relative position of different groups of observations in principal component space.

The plotting code here includes examples of: This brief exploration gave me chance to try out the new syntax in dplyr 1.0.0 and to create some explanatory figures that can help us interpret the typical scatterplots often made for the first two dimensions in a PCA.

Rearranging factor levels within groups.

Extracting values from compound variables using regular expressions.

Placing embedded subheaders into their own variables.

The code in this post tackles many small issues that come up when getting a dataset ready for analysis and/or visualization. This post describes the various steps for reading and cleaning the data, followed by some visualization of the main outputs from a dimension reduction via Principal Component Analysis (PCA). It’s a great reference, but I wanted to explore the data further. While trying to understand why my friends beat me in Age of Empires II so often, I came across this

Age of empires 2 petard