Mapping opioid overdose calls in Tempe

For this exercise, you should create the R code needed to make a density map of calls to emergency medical services for suspected opioid overdoses in Tempe, Arizona. Save your map as a JPEG image 2000 pixels wide and 2000 pixels high.

Before you start

Complete Chapters 1–7 of Learn Crime Mapping with R.

In this exercise you will practise downloading and reading data, filtering dates, converting coordinates into a spatial object, kernel density estimation, clipping datasets, adding context to a map and saving a map to a file.

What your map should show

Imagine that the City of Tempe wants to train first responders in the most-affected parts of the city to revive people suffering from opioid overdoses. To help the city decide where to prioritise training, create a map showing where calls for suspected opioid overdoses have been most concentrated since 1 January 2020.

You can download the calls data from:

https://mpjashby.github.io/crimemappingdata/tempe_opioid_calls.csv

The calls dataset documentation describes the columns and source. Filter the incident_date column to keep calls on or after 1 January 2020. These are calls for suspected overdoses, not confirmed deaths. The original coordinates were rounded to protect privacy and are approximate, so do not interpret them as exact addresses.

Unlike many cities, Tempe doesn’t have local council wards or other subdivisions. Instead, the city is divided into character areas, clusters of similar neighbourhoods. You can download the boundaries of the character areas from:

https://mpjashby.github.io/crimemappingdata/tempe_character_areas.gml

More details about the character areas data are available in the boundary dataset documentation.

Both datasets use EPSG:4326. Transform both layers to EPSG:6404, which uses metres and is suitable for analysing this part of Arizona. Acknowledge the City of Tempe as the source of both datasets under a Creative Commons Attribution licence.

Save your script in the Positron workspace you set up for the book. Store the original downloads in data/raw and the finished map in output, creating the folders if necessary. Leave the original downloaded files unchanged.

Your map should include:

  1. a density layer showing concentrations of calls since 1 January 2020, clipped to the city boundary,
  2. a suitable base map and the character areas to help readers locate the concentrations,
  3. a title, legend and supporting information explaining what the map shows, the period covered and the data sources.

Every step of your code should be accompanied by an explanatory comment (on one or more lines starting with a # symbol and a space). Your script should only produce and save one map (and possibly some R messages or warnings), not output anything else (e.g. a copy of the data).

Checklist for this exercise

Use this checklist to make sure you have done everything required for this exercise.

The map

  • The density layer includes only calls on or after 1 January 2020.
  • The density layer is clipped to the City of Tempe boundary.
  • The cell size and bandwidth make it easy to see concentrations of calls.
  • The base map and boundaries help readers identify places without distracting from the density layer.
  • The title and legend explain what is being mapped, and the map states the period covered.
  • The caption acknowledges the incident data, boundary data and base map sources.
  • The map is saved as a JPEG image 2,000 pixels wide and 2,000 pixels high.

The code

  • The necessary packages are loaded at the start of the script in a single call to pacman::p_load().
  • The code uses request() and req_perform() to download the data to local files in data/raw.
  • The code uses here() for paths within the workspace and reads the local downloads.
  • Column names and object names use snake_case.
  • The calls are converted to an SF object using their longitude and latitude coordinates.
  • Both spatial layers are transformed to EPSG:6404 before calculating and clipping the density layer.
  • The code filters the calls by date and uses |> to link data-wrangling steps.
  • The map is stored in an object and saved in output using ggsave().
  • Every step has an explanatory comment, and the script contains no temporary code such as head() or install.packages().

Check your answer

After attempting the task, read the specimen answer and explanation to see one possible solution.