Visualizing spatial temporal trends in ICE detention data

ice_spatial_temporal_trends.R

# download data from https://www.ice.gov/detain/detention-management
# see blog for more details https://jef.works/blog/2025/07/10/analyzing-ice-detention-data/

# read in data
library(readxl)
fy21 = readxl::read_xlsx('FY21-detentionstats.xlsx', sheet=4, skip=6)
fy22 = readxl::read_xlsx('FY22-detentionstats.xlsx', sheet=7, skip=6)
fy23 = readxl::read_xlsx('FY23_detentionStats.xlsx', sheet=8, skip=5)
fy24 = readxl::read_xlsx('FY24_detentionStats.xlsx', sheet=8, skip=6)
fy25 = readxl::read_xlsx('FY25_detentionStats06202025.xlsx', sheet=7, skip=6)
head(fy25)

# combine into one list
fylist <- list(fy21, fy22, fy23, fy24, fy25)
names(fylist) <- 2021:2025

library(zipcodeR)
library(dplyr)
library(tidyr)
# get data we want
alldata <- do.call(rbind, lapply(2021:2025, function(year) {
  fytest <- fylist[[as.character(year)]]
  
  # get facility names
  facs <- na.omit(unique(fytest$Name))
  
  # calculate ratio of criminal to non-criminal
  facratio <- sapply(facs, function(fac) {
    (sum(fytest[fytest$Name == fac, "Male Non-Crim"], na.rm=TRUE)+1)/
      (sum(fytest[fytest$Name == fac, "Male Crim"], na.rm=TRUE)+1)
  })
  names(facratio) <- facs
  # take a log2 (fold change)
  facratio <- log2(facratio)
  
  # where are these facilities based on zip code
  zip <- as.character(fytest$Zip)
  names(zip) <- fytest$Name
  
  # keep track of size as total detainees
  size <- rowSums(fytest[, c("Male Non-Crim", "Male Crim", "Female Non-Crim", "Female Crim")])
  names(size) <- fytest$Name
  
  # make into data frame
  zipdata <- data.frame(
    name = names(facratio),
    zip = zip[names(facratio)],
    size = size[names(facratio)],
    ratio = facratio
  )
  
  # use zipcodeR to get lat/lon
  zipcoords <- reverse_zipcode(zipdata$zip) %>%
    select(zipcode, lat, lng) %>%
    rename(zip = zipcode)
  zipdata <- left_join(zipdata, zipcoords, by = "zip") %>% drop_na()
  zipdata <- zipdata[order(zipdata$ratio),]
  
  # keep track of year and return
  zipdata$year = year
  return(zipdata)
  
}))

library(ggplot2)
library(scales)
library(maps)
library(gganimate)

# get map to visualize on US map
us_map <- map_data("state")

# plot using ggplot and gganimate
gg <- ggplot() +
  geom_polygon(data = us_map, aes(x = long, y = lat, group = group), fill = "gray95", color = "white") +
  geom_point(data = alldata, aes(x = lng, y = lat, size=size, col=ratio, group=name), alpha = 0.7) +
  scale_color_gradient2(low = 'blue', mid = 'grey', high='red', limits=c(-2,2), oob=squish) + # squish extreme values
  scale_size_binned() +
  coord_quickmap(xlim = c(-125, -66), ylim = c(25, 50)) + ## restrict to continental US
  theme_void(base_size = 10) +
  theme(plot.title = element_text(hjust = 0.5, size=16),
        legend.position="bottom", 
        legend.box = "vertical", # stack legend on top of each other
        legend.key.width = unit(1, "cm"),
        plot.margin = margin(1,2,2,1, "cm")) +
  labs(title = "ICE Detention Facilities Across the USA in {closest_state}",
       x = "Longitude", y = "Latitude", 
       color = "Log2(Male Non-Crim / Male Crim) Detainees", 
       size = "Number of Total Detainees") + 
  transition_states(year, transition_length = 1, state_length = 3) + 
  enter_fade() +
  exit_fade()
animate(gg, width=600)