In benmarwick/odmp: Officer Down Memorial Page Data

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "readme-figs/",
  out.width = "100%"
)

odmp

The goal of odmp is to make readily available the data collected by Officer Down Memorial Page, for exploration, visualisation, and analysis.

The Officer Down Memorial Page, Inc., (ODMP) is a non-profit organization dedicated to honoring America's law enforcement that have died in the course of duty. The data are the property of the Officer Down Memorial Page, Inc. We have made the data into this R package for non-commercial educational purposes, such as for use with teaching, research, criticism, commentary, and news reporting.

Installation

You can install the development version of odmp from GitHub with:

if (!require(devtools)) install.packages("devtools")
devtools::install_github("benmarwick/odmp")

A quick look at the data

We can load the data like this:

library(odmp)
data("odmp_1791_2018")

And we can inspect the data frame like this:

library(tidyverse)

# what is the structure of the data?
glimpse(odmp_1791_2018)

Here is a map of the data. This code will generate an interactive, zoomable map. Here we show just a screenshot of this map.

library(mapview)
library(sf)

# make a spatial features object for mapping
keywords_df_clean_sf <-
  odmp_age_gender_names_locations %>%
  filter(!is.na(lat)) %>%
  filter(!is.na(long)) %>%
  st_as_sf(coords = c("long", "lat"))

library(leaflet)

m <- leaflet(keywords_df_clean_sf) %>%
  addTiles() %>%  # Add default OpenStreetMap map tiles
  addMarkers(clusterOptions = markerClusterOptions(),
             popup = paste("Name:", keywords_df_clean_sf$name_and_rank, "<br>",
                           "Age:", keywords_df_clean_sf$Age, "<br>",
                           "Gender:", keywords_df_clean_sf$gender, "<br>",
                           "EOW:", keywords_df_clean_sf$eow, "<br>",
                           "Tour:", keywords_df_clean_sf$Tour, "<br>",
                           "Cause:", keywords_df_clean_sf$Cause, "<br>",
                           "Offender:", keywords_df_clean_sf$Offender))
m  # Print the map

We can make a simple infographic-style visualisaton:

library(tidyverse)
theme_set(theme_minimal(base_size = 8))
odmp <- odmp_1791_2018

# how many per year?
n_per_year <-
  ggplot(odmp,
         aes(year)) +
  geom_bar() +
  geom_text(data = odmp %>% count(year, sort = TRUE) %>% slice(1),
            aes(year,
                n + 10,
                label = paste0("Maximum in ", year),
                hjust = 1)) +
  geom_vline(data = odmp %>% count(year, sort = TRUE) %>% slice(1),
                       aes(xintercept = year),
             colour = "red") +
  ylab("Police deaths per year")

# what ranks?
n_ranks <-
  odmp %>%
  group_by(rank) %>%
  tally(sort = TRUE) %>%
  filter(!is.na(rank)) %>%
  slice(1:10) %>%
  ggplot(aes(reorder(rank, n),
             n)) +
  geom_col()  +
  coord_flip() +
  xlab("")

# what cause?
n_cause <-
  odmp %>%
  group_by(Cause) %>%
  tally(sort = TRUE) %>%
  filter(!is.na(Cause)) %>%
  slice(1:20) %>%
  ggplot(aes(reorder(Cause, n),
             n)) +
  geom_col()  +
  coord_flip() +
  xlab("")

# where are the most and least?
library(statebins)
library(viridis)
n_where <-
  odmp %>%
  group_by(State) %>%
  tally() %>%
  mutate(State = tools::toTitleCase(State)) %>%
  filter(!is.na(State), State != "NA") %>%
  ggplot(aes(state=State,
             fill=n)) +
  geom_statebins(radius = unit(0.5, "npc")) +
  coord_equal() +
  scale_fill_viridis() +
  xlab(paste0("Total number of Police deaths by state, ",
              min(odmp$year, na.rm = TRUE),
              " - ",
              max(odmp$year, na.rm = TRUE))) +
  theme_statebins()

# proportion of deaths that are heart attacks over time
prop_heart_attack <-
  odmp %>%
  mutate(Cause = if_else(Cause == 'heart attack',
                         Cause,
                         'other')) %>%
  group_by(year,
           Cause) %>%
  tally() %>%
  mutate(prop = prop.table(n)) %>%
  filter(Cause == 'heart attack') %>%
  ggplot(aes(year,
             prop,
             group = 1)) +
  geom_point(size = 2) +
  ylab("Proportion of Police deaths\neach year due to heart attack")

# devtools::install_github("thomasp85/patchwork")
library(patchwork)
p_full <-
  ( n_ranks | n_cause | prop_heart_attack) /
  ( n_per_year | n_where )

p_full