This vignette tries to implement the SPSS code used by the Murder Accountability Project to look for serial killers.

Load libraries

library(murderdata)
library(dplyr)
library(magrittr)

Get the data

We are using the supplementary data set since it has the most detal.

murder_data <- get_murderdata("supplementary")

Do some data summarization

We will create a data set that summarizes the data for us. We will group by the sex of the victim, the country of the murder and the weapon used. Then we will summarize total murders, cases solved and the percentage solved.

serial_killers <- murder_data %>%
  mutate(Solved = Solved %>% factor() %>% as.numeric() %>% subtract(1)) %>%
  group_by(VicSex, CNTYFIPS, Weapon) %>%
  summarise(total = n(),
            solved = sum(Solved),
            percentage = mean(Solved)) %>%
  ungroup() 

Possible uncaught serial killers

We want to find possible serial killers. They are (as far as I know) usually men who target women. So we will focus on female victims only, then find counties where more than 33% of a specific type of murder (by choice of weapon) has gone unsolved.

possible_serial_killers <- serial_killers %>%
  filter(VicSex == "Female",
         percentage <= .33) %>%
  mutate(unsolved = total - solved) %>%
  arrange(desc(unsolved))

Lets have a look

possible_serial_killers %>% select(-VicSex) %>% head() %>% knitr::kable()

CNTYFIPS	Weapon	total	solved	percentage	unsolved
St. Louis city, MO	Firearm, type not stated	203	62	0.3054187	141
Fulton, GA	Strangulation - hanging	133	33	0.2481203	100
King, WA	Other or type unknown	108	31	0.2870370	77
St. Louis city, MO	Other or type unknown	107	35	0.3271028	72
District of Columbia	Strangulation - hanging	71	19	0.2676056	52
Jackson, MO	Strangulation - hanging	73	22	0.3013699	51

Here we see that in Fulton County in Georgia we have over a 133 strangled female victims, but only 33 of them have been solved… Is there a potential serial killer lurking in Fulton County?