Crime Mapping and Spatial Data Analysis using R

This chapter aims to focus on introducing good practice in map design and presentation. When putting a map together, you need to think about its intended audience (their level of expertise, whether you want them to interact with the map), purpose, and format of delivery (e.g., printed, web, projected in a screen, etc.). There are many design decisions you need to consider: fonts, labels, colour, legends, layout, etc. In this chapter we provide a general introduction to some basic design principles for map production. These themes, and the appropriate election of symbol representation, are the subject matter of cartography, the art and science of map making. Within cartography, a considerable body of research and scholarship has focused on studying the visual and psychological implications of our mapping choices. As noted in previous chapters, one of the problems with maps is that powerful as a tool as they can be, they can lead to misunderstanding. What the mapmaker chooses to emphasise and what the map reader sees may not be the same thing. We will work you through an example of a fairly basic map and the process of taking to a point where it could be ready for presentation to an audience other than yourself.

In this chapter we will be working with some data published by Hungarian police available online ([144]). Specifically we will be looking at some statistics related to drink driving. Drink driving is one of a number of problems police confront that relate to impaired and dangerous driving. Hungary has a strict drink driving policy, with the maximum drink diving limit being 0.0 BAC. Most European countries are at 0.5 BAC, while the UK is 0.8 (except 0.5 for Scotland). We have records for each county with the number of breathalyser checks carried out, and the number of these which returned a positive result.

We have downloaded and saved these data in the data folder made available with this textbook. The two data sets we will use are this drink driving data set, called drink_driving.csv, and a geometry of the counties within Hungary called hungary.geojson.

In this chapter, we will be making use of the following libraries:

# Packages for reading data and data carpentry
library(readr)
library(dplyr)

# Packages for handling spatial data and for geospatial carpentry
library(sf)
library(ggspatial)
library(rnaturalearth)

# Packages for mapping and visualisation
library(ggplot2)
library(RColorBrewer)
library(ggrepel)
library(cowplot)

So let’s read in our datasets, and join the attribute data to the geometry using left_join() (if you’re unsure about any of the below code, revisit Chapter 1 of this book for a refresher!).

# read in geojson polygon for Hungary
hungary <- st_read("data/hungary.geojson")

#read in drink driving data
drink_driving <- read_csv("data/drink_driving.csv")

#join the csv (attribute) data to the polygons
hu_dd <- left_join(hungary, drink_driving, by = c("name" = "name"))

We can now use this example to talk through the important principles of good visualisation of spatial data. We draw specifically from two areas of research: cartography and data visualisation. Let’s start with cartography.

Cartographers have always been concerned about the appearance of maps and how the display marries form with function ([145]). As there is no definitive definition for what is meant by cartographic design, it can be challenging to evaluate what makes good design. However, there are themes and elements which can be used to guide the map maker, and offer points of reflection to encourage thoughtful designs.

The primary aim of maps is the communication of information in an honest and ethical way. This means each map should have a clear goal and know its audience, show all relevant data and not use the data to lie or mislead ([146]). It should also be reproducible, transparent, cite all data sources, and consider diversity in its audience ([146]). So what does that mean for specifically implementing these into practice. While a good amount of critical thought from the map-maker will be required, there are aids we can rely upon. For example, [147] developed a map evaluation checklist which asks the map maker a series of questions to guide their map-making process. The questions fall into three broad categories:

We will discuss these elements in this chapter to some degree, and the recommended reading will guide the reader to further advice on these topics.

Data visualisation is a somewhat newer field; however, it seems to encompass the same guiding principles when considering what makes good design. According to [148] three principles offer a guide when deciding what makes a good data visualisation. It must be: trustworthy, accessible, and elegant. The first principle of trust speaks to the integrity, accuracy, and legitimacy of any data visualisation we produce. [148] suggests this principle to be held above all else, as our primary goal is to communicate truth (as far as we know it) and avoid at all cost to present what we know to be misleading content (see as well [149]). Accessibility refers to our visualisation being useful, understandable, and unobtrusive, as well as accessible for all users. There are many things to consider in your audience such as dynamic of need (do they have to engage with your visualisation, or is it voluntary?), subject-matter knowledge (are they experts in the area, are they lay people to whom you must communicate a complex message? and many other factors) (see [148]). Finally, elegance refers to aesthetics, attention to detail, and an element of doing as little design as possible — meaning a certain invisibility whereby the viewer of your visualisation focuses on the content; rather than the design — that is the main point is the message that you are trying to communicate with your data. There are various schools of thought within data visualisation research. For example, the work of [4] emphasises clean, minimalist approaches to data visualisation, emphasising a low data-to-ink-ratio, which means all the ink needed to print the visualisation should contribute data to the graph. However, other research has explored the usefulness of additional embellishments on charts (which Tufte calls "chart junk") — finding there may be value to these sorts of approaches as well (see [150]). In this chapter, we will aim to bring together the above principles, and work through a practical example of how to apply these to the maps we make.