Visualizing Spatial Data via Linked Micromaps

Linked micromap plots display geospatial data on small maps linked by color to statistical graphics, enabling the user to explore and to communicate the underlying patterns in the data. Sorting by a statistical column can reveal geospatial clusters of similar values or correlations between the columns.

Carr et al. added small maps to row-labeled plots to show the spatial context of the data. Implementations of the plots were first in S-plus, then JAVA and then an R package, micromapST, first released in 2013. We designed the R package to produce publishable quality output with minimal user input. By using mostly basic R functions, iterative applications for exploratory spatial data analysis run quickly.

Although the first version only displayed U.S. maps at the state level, subsequent versions can display data for other geographic units. The new version allows users to supply their own geographic boundary file so that linked micromaps can be produced for virtually any geospatial structure. Also, new types of glyphs and user options have been added.

We will describe linked micromap plots and demonstrate how to use the latest version of micromapST to produce them
 

The linked micromaps approach was originally developed as an improvement to choropleth maps for displaying statistical summaries connected with spatial areal units, such as countries, states, and counties. Two R packages to create linked micromaps were published in 2015. These are the micromap and micromapST packages. The latter is for data indexed to the 50 US states and DC. The package handles the formatting needed for the micromap package and offers several options for statistical displays (scatterplots, boxplots, time series plots, and more). The micromapST package is useful, but it can be problematic specifying the data frames needed to specify the desired graphic. Furthermore, exploring data through visualization is easier, faster, and more intuitive using a graphical user interface. This is the motivation behind the R Shiny micromapST app, which is the topic of this talk. This presentation will serve as a tutorial and introduction to micromapST and the Shiny app using real-world data and applications. 

Only a few stand-alone attempts have been made in the past to create linked micromap plots
for point locations such as climate stations and cities, rather than for areal locations (i.e., polygons).
This requires that the point location is extended to a small (circular or quadratic)
area that can be color-coded in a linked micromap plot. In this talk, we provide an
overview of necessary steps to produce linked micromap plots for point locations
to create areas of suitable sizes and to avoid overplotting of nearby point locations.
A series of examples will highlight the benefits of linked micromap plots for point locations.