gpsmartr

The goal of gpsmartr is to illustrate the geographic profiling method described in Curtis-Ham et al (2022), the Geographic Profiling: Suspect Mapping and Ranking Technique (GP-SMART). The goal of GP-SMART is to support geographic profiling in police investigations, by mapping and ranking suspects for an input crime, based on the location and attributes of the suspects’ activity locations (nodes). Users should read that paper before using gpsmartr. Its operational use is at the user’s own risk; accuracy outside the test data used to develop the method (as described in the paper) is not guaranteed.

The functions in gpsmartr require the user to have a) an input crime and b) a dataset of suspect activity nodes. These files need to include specific variables describing the location, time and other attributes of both the input crime and suspect activity nodes. Examples of both files are provided as package data.

The functions should be run in the order shown in the example below. They prepare the input data, run the GP-SMART process, then map the output.

Reference

Curtis-Ham S., Bernasco, W., Medvedev, O. N., & Polaschek, D. L. L (2022). ‘A new geographic profiling method for mapping and ranking suspects in crime investigations: GP-SMART’. Journal of Investigative Psychology and Offender Profiling. https://doi.org/10.1002/jip.1585

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("Sophie-c-h/gpsmartr")

Example

This example steps through the gpsmartr workflow:

First prepare the input crime and input suspect activity location (node) data. These functions will return an error if the input datasets do not include the necessary columns in the necessary format. This example uses the built in package data, which is fictional data that approximates the real crime and suspect data used to develop and test GP-SMART.

library(gpsmartr)

data(example_input_crime_raw)
data(example_input_suspects_raw)
input_crime <- fn_prepare_input_crime(example_input_crime_raw)
input_suspects <- fn_prepare_suspect_data(example_input_suspects_raw)

Then run the GP-SMART process to generate a shortlist of ranked suspects for the input crime, based on the suspects’ activity locations.

gpsmart_output <- fn_gpsmart(
   input_crime = input_crime,
   input_suspects = input_suspects,
   search_radius = 10,
   return_node_predictions = TRUE # set this to false if not wanting to map the output (next step)
 )

Then map the output to explore it interactively.

map <- fn_map_gpsmart_output(
   gpsmart_output,
   coordinate_system = 2193
 )