Page 292 - Contributed Paper Session (CPS) - Volume 4
P. 292
CPS2230 Braden P. et al.
Using ‘rmapshaper’ to Modify Boundary Files for
Use in Linked Micromap Plots
Braden Probst, Jürgen Symanzik
Utah State University, Department of Mathematics and Statistics, Logan, UT, USA
Abstract
Linked micromap plots have been in use since their creation in the 1990’s.
Initially, the underlying code was complex and the shapefiles used to represent
the spatial boundaries were not easily obtained or efficient to use. Using
modern software, the process of modifying and simplifying shapefiles has
become more accessible, facilitating the ability to more easily create and
analyze linked micromap plots --- and doing so on a larger scale.
Keywords
Spatial Data, Data Visualization, Map, Shapefile
1. Introduction
Over the last three decades, linked micromap plots (LMplots) have been
developed as a way to visualize potential trends within some spatial
geographic data and within some associated statistical variable(s) [6]. In the
statistical software environment R [4], LMplots can be created via the
‘micromap’ R package [3] for example.
While there exist different plot types that tie statistical data to spatial data,
such as choropleth maps, LMplots employ some visualization concepts that,
in many cases, make the statistical trends and spatial patterns more visible and
clear. The first of these concepts used by LMplots that becomes one of its
greatest strengths, is the concept of small multiples. The benefit here is rather
than having all of the data shown in a single (large) map, the data are spread
into several smaller and comparable maps.
The basic structure of LMplots is an array of several vertical panels with the
underlying data being oriented by row. Typically, although not exclusively, the
first of these panels is used to plot spatial data in the form of a shapefile for
the geographic regions of interest, along with the labels and names of these
regions in panels three and four. One or more columns of statistical data
corresponding to the regions in the spatial panel are included in the
subsequent panels. The panels plotting the statistical variables can take on a
variety of forms, although dot plots and line plots seem to be the most
common choices. The entire LMplot is then sorted by one of the variables that
were included. The sorting adjusts all of the rows according to the specific
variable and direction of the sorting. The complete LMplot shows whether
281 | I S I W S C 2 0 1 9