Plot violent crime trends from Omaha PD 2016-2021 and compare to 2022

omaha_violent_crime_trends.R

library(tidyverse)
library(lubridate)
library(janitor)

# get urls of csv incident files to download for year 2015 - 2022
urls <- map_chr(2015:2022, ~ paste0("https://police-static.cityofomaha.org/crime-data/",
                                    .x, "/Incidents_", .x, ".csv"))

# download all years, combine into one df and clean up variable names
incidents <- read_csv(urls) |>
  clean_names()

# deduplicate incidents, create month, year vars, new var for combining violent crimes
incidents_clean <- incidents |>
  distinct(rb_number, reported_date, reported_time,
           statute_ordinance_description, occurred_location, .keep_all = TRUE) |>
  mutate(
    reported_date = mdy(reported_date),
    year = year(reported_date),
    month = month(reported_date),
    violent_crime = case_when(
      statute_ordinance_description == "HOMICIDE" ~ "Homicides",
      statute_ordinance_description %in% c("ASSAULT - FELONY - AGGRAVATED",
                                           "ASSAULT - FELONY - AGGRAVATED -ATTEMPT") ~ "Agg Assaults and Attemped Agg Assaults",
      str_detect(statute_ordinance_description, "ROBBERY") ~ "Robberies and Attempted Robberies",
    )
  ) |>
  filter(year >= 2016)

# for 2022, count number of violent crimes by month
violent_crime_2022 <- incidents_clean |>
  filter(!is.na(violent_crime), year == 2022, month <= 8) |>
  count(violent_crime, month)

# for each month from 2016-2021, calc mean, min, and max of each violent crime
by_month <- incidents_clean |>
  filter(!is.na(violent_crime), year <= 2021) |>
  count(violent_crime, year, month) |>
  group_by(violent_crime, month) |>
  summarize(
    mean = mean(n),
    min = min(n),
    max = max(n)
  ) |>
  ungroup()


# plot - line for 2022, dashed line for 2016-21 mean and filled range for 2016-21 min/max
by_month |>
  ggplot(aes(month)) +
  geom_ribbon(aes(ymin = min, ymax = max, fill = "c"), alpha = 0.1) +
  geom_line(aes(y = mean, color = "a"), linetype = "dashed") +
  geom_line(data = violent_crime_2022, aes(y = n, color = "b"), lwd = 1) +
  scale_x_continuous(breaks = 1:12) +
  scale_fill_manual(labels = "Min/Max 2016-21", values = c(c = "black")) +
  scale_color_manual(labels = c("Mean 2016-21", "2022"),
                     values = c(a = "black", b = "red")) +
  ylim(0, NA) +
  facet_wrap(vars(violent_crime), scales = "free", nrow = 2) +
  labs(
    x = "Month",
    y = NULL,
    color = NULL,
    fill = NULL
  ) +
  theme_minimal() +
  theme(
    plot.margin = margin(20, 20, 20, 20),
    panel.grid.minor = element_blank(),
    legend.position = c(0.8, 0.2)
    )

This is an R replication of Justin Nix's Stata code to see how a recent month of crime data compares to prior years and create the following plot.