Program Profile: Predictive Policing Model in Los Angeles, Calif.

This program involved the use of a crime forecasting model to direct police patrol to dynamic hot spots to reduce crime. The program is rated Promising. There were statistically significant reductions in daily crime volume (specifically burglary, car theft, and burglary-theft from vehicle) for police patrols in the treatment condition, compared with patrols in the control condition.

A Promising rating implies that implementing the program may result in the intended outcome(s).

This program's rating is based on evidence that includes at least one high-quality randomized controlled trial.

A Promising rating implies that implementing the program may result in the intended outcome(s).

This program's rating is based on evidence that includes at least one high-quality randomized controlled trial.

Program Goals

The hot spot policing approach deploys policing resources in small geographic areas in which crime is most heavily concentrated (Sherman and Weisburd 1995). Predictive policing is intended to deter crime by selecting hot spots with the highest concentrations of crime as the targets for police patrol. Predictive policing involves the use of advanced statistical methods to identify actionable patterns from crime data, allowing police services to predict and anticipate future crime events (Hardyns and Rummens 2018). Specifically, the epidemic-type aftershock sequence model (Mohler et al. 2011; Mohler 2014) was a predictive policing software that estimated the risk associated with both long-term hotspots and short-term models of near-repeat risk. The goal of predictive policing was to help departments determine where to allocate police resources to reduce crime most effectively and efficiently.

Program Components

The predictive policing algorithm “epidemic-type aftershock sequence” model was premised on the idea that crimes cluster spatially due, in part, to “near repeat” or contagion effects. Locations that previously experienced crime were likely to experience crime again, though at a potentially less severe level (Mohler et al. 2015; Ensign et al. 2018). The model used crime incident data only, including incidents discovered by police and reported by residents.

The model was based on four inputs: date, time, location, and type of crime. These inputs were submitted to a machine-learning algorithm to generate crime risk predictions mapped on a grid with a single cell size of 150 meters by 150 meters. Different crime types were displayed on the same map, and the system allowed time and spatial range to be chosen by the user. Generally, a new map was generated for each police patrol shift because the model used the most-recent crime data available in a department’s live records management system. The generated maps could also be sent to mobile devices, which police officers could use in the field (Hardyns and Rummens 2018). Officers were deployed in areas with the highest predicted concentrations of crime. Any new crime incidents discovered in those locations were then fed back into the system (Ensign et al. 2018).

Study

Mohler and colleagues (2015) conducted a single-blind, randomized, controlled trial in three divisions of the Los Angeles Police Department (LAPD) in California to assess the effects of a fully automated statistical algorithm for predictive policing, the epidemic-type aftershock sequence (ETAS) model, on crime volume in dynamic hot spots.

The field trials were conducted in three LAPD divisions on the following dates: Foothill Division from Nov. 7, 2011, to April 27, 2012; North Hollywood Division from March 31, 2012, to Sept. 14, 2012; and Southwest Division from May 16, 2012, to Jan. 10, 2013. Rather than dividing patrol areas into partially matched blocks, patrol “mission maps” were generated independently by a crime analyst (the control condition) and the ETAS algorithm (the treatment condition) each day. Mission maps were identical in outward appearance, except for the exact placement of prediction boxes. Only after each set of predictions was generated was it randomly determined, using a Bernoulli random number generator, whether the control or treatment condition would be deployed to the field for the next 24-hour period. Hot spot locations dynamically changed each day, therefore days were randomly assigned to the treatment or control conditions. Command staff, supervisors, and patrol officers in each of the deployment areas were not aware of the distinction between treatment and control mission maps, and the police officers involved were the same on both control and treatment days. Field-deployed missions consisted of twenty 150-by-150-meter prediction boxes (which is the size of a city block in Foothill and held constant across the experimental regions) for each 12-hour shift per division. Patrol officers were directed to use available time to “get in the box” and police what they saw; therefore, control and treatment conditions compare directed patrol patterns in space and time, not differences in field tactics. A total of 62 control days and 62 treatment days in Foothill, 82 control days and 70 treatment days in North Hollywood, and 117 control and 117 treatment days in Southwest were randomly assigned during the trial, for a total of 510 days.

In the treatment condition, an ETAS crime forecast consisted of the top 20 prediction boxes in rank order displayed on a map and available for directed police patrol. The ETAS model was implemented as a fully automated, cloud-based machine learning system. Software was installed on the police records management system server that encrypted and sent crime report data (address, date, time, crime type) once every hour to cloud-based servers, where the data were geocoded using Google application programming interfaces and stored for later use. Once a day at 4:00 a.m. the ETAS parameters were re-estimated using the previous 365 days of crime data (up to the latest records) and boxes were ranked by intensity. Patrol commanders logged on with a web interface before each patrol shift and printed out a report that was then distributed to officers. The software also allowed the analyst on control days to place 20 boxes (identical in size and appearance to ETAS boxes) on the Google map instead of ETAS–generated boxes. The report contained a map of hot spot locations and a list of nearest cross-streets of the hot spot locations. Missions were set daily at 5:00 a.m. for two sequential shifts from 6:00 a.m. to 6:00 p.m. and 6:00 p.m. to 6:00 a.m. In Foothill, mission predictions were made for the 24-hour period starting at 4 p.m.

For the control condition, crime analysts placed a fixed number of 150-by-150-meter boxes within their operational environment in specified time windows. The goal of box placement was to identify a small set of locations where the analysts expected crime most likely to occur. In Los Angeles, three crime analysts acting independently were responsible for placing prediction boxes for Foothill, North Hollywood, and Southwest Divisions. Crime analysts were free to use any information and methods at their disposal to generate predictions. The LAPD followed a COMPSTAT policing model focused on the analysis of 7-day crime maps supplemented with ad hoc street-level intelligence. The LAPD analysts emphasized small clusters of recent crimes as signaling emerging problems in need of short-term response to interrupt formation of crime hot spots. An easy-to-use web interface was developed to allow analysts to choose prediction locations by clicking on a Google map. Analysts spent considerable time engaged in analysis of crime and intelligence data before committing to predictions. Each analyst spent about 1.5 to 2.0 hours each day inspecting crime maps and recent crime reports before placing their prediction boxes. In the Foothill Division, 20 predictions were generated for each 24-hour period beginning at 4:00 p.m. each day. In both North Hollywood and Southwest Divisions, 40 prediction boxes were generated each day, evenly divided between two shifts (6:00 a.m. to 6:00 p.m. and 6:00 p.m. to 6:00 a.m.).

Officer time on mission in control and treatment prediction boxes was used as the primary measure of officer activity. Officers generated a time-stamped call event on their in-car mobile data terminal upon entering and leaving any mission box. The total amount of time spent within prediction boxes was the officer time on mission, and each incident of an officer spending time within a box was a mission stop. Reported crimes were tabulated according to whether they occurred strictly inside or outside active control and treatment boxes. A crime occurring strictly inside an active prediction box (which is a box generated before the start of a patrol shift and held constant in that location for the duration of the shift) is considered a successful prediction. Regression models were constructed to assess the relationship between daily crime volume and patrol time on mission. Changes in crime volume (specifically burglary, car theft, and burglary-theft from vehicle) were assessed for all three LAPD divisions as a whole. No subgroup analysis was conducted.

Mohler and colleagues (2015) examined the potential cost savings from using the epidemic-type aftershock sequence model (ETAS) at mean patrol levels (31 minutes per box per day), which corresponded with 4.3 fewer crimes per week per division. The societal costs per crime calculated by McCollister, French, and Fang (2010)—broken down by costs to the victim, the police and court system, and the individual who offended (converted to 2013 dollars); the relative frequencies of grand theft auto, burglary theft from motor vehicle, and burglary; along with the average crime rate per Los Angeles Police Department division at zero patrol levels—were used to estimate the weekly cost of crime per division, without predictive policing. The authors projected a $17,258,801 annual savings to the Los Angeles Police Department if the ETAS model were to be used for 31 minutes per day in each hot spot, compared with no patrol.

The epidemic-type aftershock sequence model is known as PredPol, and recently became Geolitica. The Los Angeles Police Department no longer has a contract with PredPol (Bhuiyan 2021).

These sources were used in the development of the program profile:

These sources were used in the development of the program profile:

Bhuiyan, Johana. “LAPD Ended Predictive Policing Programs Amid Public Outcry: A New Effort Shares Many of Their Flaws.” Guardian, Nov. 8, 2021.

Ensign, Danielle, Sorelle A. Friedler, Scott Neville, Carlos Scheidegger, and Suresh Venkatasubramanian. 2018. Runaway Feedback Loops in Predictive Policing. Proceedings of Machine Learning Research 81:1–12.

Hardyns, Wim, and Anneleen Rummens. 2018. “Predictive Policing as a New Tool for Law Enforcement? Recent Developments and Challenges.” European Journal on Criminal Policy and Research 24:201–18.

McCollister, Kathryn E., Michael T. French, and Hai Fang. 2010. “The Cost of Crime to Society: New Crime-Specific Estimates for Policy and Program Evaluation.” Drug and Alcohol Dependence 108(1-2):98–109.

Mohler, George O., Martin B. Short, P. Jeffrey Brantingham, Frederic Paik Schoenberg, and George E. Tita. 2011. “Self-Exciting Point Process Modeling of Crime.” Journal of the American Statistical Association 106(493):100–108.

Mohler, George O. 2014. “Marked Point Process Hotspot Maps for Homicide and Gun Crime Prediction in Chicago.” International Journal of Forecasting 30:491–97.

PredPol. 2020. “Predictive Policing Technology.” Santa Cruz, Calif.

Sherman, Lawrence William, and David L. Weisburd. 1995. “General Deterrent Effects of Police Patrol in Crime ‘Hot Spots’: A Randomized, Controlled Trial.” Justice Quarterly 12(4):625–48.

Following are CrimeSolutions-rated programs that are related to this practice:

Hot Spots Policing

Hot spots policing strategies focus on small geographic areas or places, usually in urban settings, where crime is concentrated. Through hot spots policing strategies, law enforcement agencies can focus limited resources in areas where crime is most likely to occur. This practice is rated Effective for reducing overall crime and rated Promising for reducing violent, property, public order, and drug and alcohol offenses.

Evidence Ratings for Outcomes

	Crime & Delinquency - Multiple crime/offense types
	Crime & Delinquency - Violent offenses
	Crime & Delinquency - Property offenses
	Crime & Delinquency - Public order offenses
	Crime & Delinquency - Drug and alcohol offenses

Geographically Focused Policing Initiatives

Geographically focused policing initiatives increase the presence and visibility of police officers at specific high-crime locations to significantly reduce crime and disorder. This practice is rated Promising for reducing crime in treatment areas relative to control areas.

Evidence Ratings for Outcomes

Crime & Delinquency - Multiple crime/offense types

Why might a practice's outcome ratings differ from the ratings of specific programs encompassed by that practice?