Introduction

This analysis attempts to visualize and contextualize spatial and temporal patterns in point data of crimes in St. Louis, Missouri. The analysis was conducted using a variety of packages in the R scripting language.

Data

The St. Louis crime dataset contains 557 records which capture all reported homicides, arson, and DUIs. Each record also lists a year and month of the incident along with a handful of other fields. The data spans from August 2013 to August 2014.

Three important notes

Because the day of the incident is not captured in the data, all analysis was performed with a temporal resolution of 1 month.
January 2014 had no reported incidents. This was likely caused by a reporting issue and not an actual drop in crime.
For the month-by-month analysis, August is captured twice (2013 and 2014), so the data from that month is averaged between the two years.

Analysis

This section of the report walks through results derived from analysis in R

load relevant libraries in R

library(leaflet)
library(tidyverse)
library(sf)
library(tidyr)

set filepaths for crime data and foundational GIS layers

file_dir_crime <- "C:/temp/PSU_operational/GEOG586/Lesson2/Geog586_Les2_Project/crime/"
file_dir_gis <- "C:/temp/PSU_operational/GEOG586/Lesson2/Geog586_Les2_Project/gis/"

read in csv of crime data

stl_crime <- read.csv("C:/temp/PSU_operational/GEOG586/Lesson2/Geog586_Les2_Project/crime/crimeStLouis20132014b.csv", header = TRUE, sep=",")

take a peak at what the crime data contains

summary(stl_crime)

##      recno        crimetype               xL               yL       
##  Min.   :  235   Length:557         Min.   :-90.33   Min.   :38.54  
##  1st Qu.:14284   Class :character   1st Qu.:-90.26   1st Qu.:38.60  
##  Median :24229   Mode  :character   Median :-90.24   Median :38.65  
##  Mean   :26208                      Mean   :-90.24   Mean   :38.64  
##  3rd Qu.:39623                      3rd Qu.:-90.22   3rd Qu.:38.67  
##  Max.   :52145                      Max.   :-90.18   Max.   :38.76  
##       year          month            count    codemonth        
##  Min.   :2013   Min.   : 2.000   Min.   :1   Length:557        
##  1st Qu.:2013   1st Qu.: 4.000   1st Qu.:1   Class :character  
##  Median :2014   Median : 8.000   Median :1   Mode  :character  
##  Mean   :2014   Mean   : 6.996   Mean   :1                     
##  3rd Qu.:2014   3rd Qu.: 9.000   3rd Qu.:1                     
##  Max.   :2014   Max.   :12.000   Max.   :1                     
##      crimet          district       Neighborho   
##  Min.   : 10000   Min.   :0.000   Min.   : 0.00  
##  1st Qu.: 10000   1st Qu.:2.000   1st Qu.:18.00  
##  Median : 83000   Median :4.000   Median :50.00  
##  Mean   :106830   Mean   :4.129   Mean   :42.53  
##  3rd Qu.:211000   3rd Qu.:6.000   3rd Qu.:64.00  
##  Max.   :212000   Max.   :9.000   Max.   :83.00

explore what types of crime were recorded in the data

recorded_crimes <- unique(stl_crime$crimetype)
recorded_crimes

## [1] "homicide" "arson"    "dui"

Figure 1: exploring the macro temporal trends in the data by plotting number of incidents over time. The gap in data for January 2014 is immediately apparent. Within the scope of the dataset, crime was at its lowest non-zero rate in February 2014 and peaked in April 2014.

stl_crime <- stl_crime %>%
  mutate(date = as.Date(paste(year, month, "01", sep = "-")))

monthly_counts <- stl_crime %>%
  group_by(date) %>%
  summarise(count = n())

complete_monthly_counts <- monthly_counts %>%
  complete(date = seq.Date(min(date), max(date), by = "month")) %>%
  replace_na(list(count = 0))

complete_monthly_counts$date <- factor(complete_monthly_counts$date, levels = complete_monthly_counts$date, labels = format(complete_monthly_counts$date, "%Y-%m"))

ggplot(complete_monthly_counts, aes(x = date, y = count)) +
  geom_col() +
  labs(title = "Figure 1: Number of crimes, per month",
       x = "Month",
       y = "Crimes") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

Figure 2: exploring the temporal trends in the data by plotting number of incidents over time broken out by crime type. Again, the gap in Janaury 2014 is apparent.

crime_counts <- stl_crime %>%
  group_by(date, crimetype) %>%
  summarize(count = n()) %>%
  ungroup()

complete_monthly_counts <- crime_counts %>%
  complete(date = seq.Date(min(date), max(date), by = "month"), crimetype) %>%
  replace_na(list(count = 0))

complete_monthly_counts$date <- as.factor(format(complete_monthly_counts$date, "%Y-%m"))

ggplot(complete_monthly_counts, aes(x = date, y = count, color = crimetype, group = crimetype)) +
  geom_line() +
  labs(title = "Figure 2: Number of crimes, per month-year by crime type",
       x = "Month",
       y = "Count of Events",
       color = "Crimes") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

As Figure 2 begins to show, there is a cyclical nature to this data. Figure 3 illustrates this further showing the number of occurrences by month only. As previously discussed, Janaury has no data and August is averaged between 2013 and 2014. Within the scope of the dataset, DUIs peak in April and arson peaks in September.

monthly_counts <- stl_crime %>%
  group_by(month = format(date, "%m"), crimetype) %>%
  summarise(count = n(), .groups = 'drop') %>%
  mutate(count = if_else(month == "08", count / 2, count))

monthly_counts$month <- factor(monthly_counts$month, levels = sprintf("%02d", 1:12), labels = month.abb)

ggplot(monthly_counts, aes(x = month, y = count, color = crimetype, group = crimetype)) +
  geom_line() +
  labs(title = "Figure 3: Number of crimes, per month by crime type",
       x = "Month",
       y = "Crimes",
       color = "Crime Type") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

Shifting from temporal to spatial trends, the interactive map made using leaflet shows possible clusters and patterns in crimes visible to the naked eye. These include:

a concentration of homicides in the northern half of the city as well as the southeastern quadrant.
a similar concentration of arson in the northern half of the city as well as the southeastern quadrant.
DUIs more evenly spread across the city with some small clusters just south of downtown.

stl_boundary <- st_read(paste(file_dir_gis, "stl_boundary_ll.shp", sep = "")) %>%
  st_transform(crs = 4326) %>%
  st_make_valid()

## Reading layer `stl_boundary_ll' from data source 
##   `C:\temp\PSU_operational\GEOG586\Lesson2\Geog586_Les2_Project\gis\stl_boundary_ll.shp' 
##   using driver `ESRI Shapefile'
## Simple feature collection with 1 feature and 4 fields
## Geometry type: POLYGON
## Dimension:     XY
## Bounding box:  xmin: -90.32052 ymin: 38.532 xmax: -90.16631 ymax: 38.77435
## Geodetic CRS:  WGS 84

geocoded_data_sf <- st_as_sf(stl_crime, coords = c("xL", "yL"), crs = 4326)

homicide <- filter(geocoded_data_sf, crimetype == "homicide")
arson <- filter(geocoded_data_sf, crimetype == "arson")
dui <- filter(geocoded_data_sf, crimetype == "dui")

leaflet() %>%
  addTiles() %>%
  addPolygons(data = stl_boundary, color = "black", fillColor = "transparent") %>%
  addCircleMarkers(data = homicide, group = "Homicide", 
                   radius = 5, color = "red", fillOpacity = 0.5,
                   popup = ~paste("Month: ", geocoded_data_sf$month, "<br>",
                                  "Year: ", geocoded_data_sf$year, "<br>",
                                  "District: ", geocoded_data_sf$district, "<br>",
                                  "Neighborhood: ", geocoded_data_sf$Neighborho)) %>%
  addCircleMarkers(data = arson, group = "Arson", 
                   radius = 5, color = "blue", fillOpacity = 0.5,
                   popup = ~paste("Month: ", geocoded_data_sf$month, "<br>",
                                  "Year: ", geocoded_data_sf$year, "<br>",
                                  "District: ", geocoded_data_sf$district, "<br>",
                                  "Neighborhood: ", geocoded_data_sf$Neighborho)) %>%
  addCircleMarkers(data = dui, group = "DUI", 
                   radius = 5, color = "green", fillOpacity = 0.5,
                   popup = ~paste("Month: ", geocoded_data_sf$month, "<br>",
                                  "Year: ", geocoded_data_sf$year, "<br>",
                                  "District: ", geocoded_data_sf$district, "<br>",
                                  "Neighborhood: ", geocoded_data_sf$Neighborho)) %>%
  # Add layer control
  addLayersControl(overlayGroups = c("Homicide", "Arson", "DUI"),
                   options = layersControlOptions(collapsed = FALSE)) %>%
  addLegend(position = "bottomright",  # Adjust position as needed
            colors = c("black", "red", "blue", "green"),
            labels = c("City boundary", "Homicide", "Arson", "DUI"), 
            opacity = 1)

Additional Analytic Opportunities

Beyond what was covered in this brief report, additional insights could be gleaned by:

incorporating day and time of day into temporal analysis
analyzing a complete years worth of data without gaps
looking for patterns across wards, police districts, etc. (while also keeping the modifiable areal unit problem in mind)
carrying out additional exploration and quantification of locations with high and low crime counts.

neighborhood_counts <- stl_crime %>%
  group_by(Neighborho) %>%
  summarise(count = n()) %>%
  ungroup()

avg_crime <- sum(neighborhood_counts$count)/88
avg_crime

## [1] 6.329545

Conclusion

This analysis showed distinct spatial and temporal patterns in St. Louis’ crime data for the included years.

St. Louis Crime Analysis

Kyle Devine

2024-05-27