Machine learning integration for crime busting in Malaysia

Machine learning integration for crime busting in Malaysia

SHARE

Smart City method of crime busting has been introduced in Malaysia through its National Key Result Area (NKRA) Safe City Programme. Research is currently on to predict the next crime location, of a wanted serial criminal suspect in particular, using integration of a centralised GIS and machine learning.

Safe City Monitoring System (SCMS) is an initiative under the National Key Result Area (NKRA) Safe City Programme, introduced by Malaysian Government in 2010. Utilising the mapping of crime on a GIS platform, Minister of Housing and Local Government Malaysia has launched SCMS in April 2011. Deploying an ingenious way to quickly visualise the hot spots of crimes, SCMS makes good use of GIS in collaboration with the existing Police Reporting System through collaboration with the Royal Malaysia Police (RMP).

Through this collaboration, Federal Town and Country Planning Department (FTCPD) takes custodianship of the GIS infrastructure, spatial data and application while Royal Malaysia Police takes custodian to the crime data. When a police report is made by the public, the attending officer plots the location of crime as well as keying-in the necessary information into the Police Reporting System. Table below lists the types of crime covered by the SCMS.

 

Categories of Crime Mapped by Safe City Programme

 Category

 Type

 Violent Crime

 Murder

 Attempted Murder

 Gang robbery with firearm

 Gang robbery without firearm

 Robbery with firearm

 Robbery without firearm

 Rape

 Inflicting injury

 Property Crime

 House break-in and theft (day)

 House break-in and theft (night)

 Lorry / van theft

 Car theft

 Motorcycle theft

 Snatch theft

 Other thefts

(Source: FTCPD 2008)

 

Revolutionising Police Reporting System

With cooperation from Royal Malaysia Police and the use of GIS, the Police Reporting System is revolutionised by the way of how crimes are recorded and later visualised for the purpose of busting crime. The effective execution of SCMS is reliant on GIS Crime Mapping as the key measure.

Figure below shows the crime mapping process. As an incident happened, a police report is made by the victim. Scene and incident information are recorded and the location of the incident is plotted. The location of the incident is then verified by the investigating officer.

 Crime Mapping Process (Source: FTCPD 2011)

 

An existing geospatial database then gets updated with the crime incident and its location as per reported into the Police Reporting System. This integration between Federal Town and Country Planning Department’s GIS facility and Royal Malaysia Police’s Reporting System puts ease and automation to the crime maps and hot spots visualisation hence becoming an important enabler to the Crime Mapping database.

 

Visualisation of Crime Locations in Klang Valley area (Source: FTCPD 2010)

 

Figure above shows the selection of Kajang, Subang Jaya, Klang, Petaling Jaya, Selayang, Ampang Jaya and Shah Alam municipalities as well as the Federal Territory of Kuala Lumpur illustrating the focus of a selected crime type. Figure below visualises crime hot spot analysis. Electronic maps easily allows a multitude of capabilities such as adding or removing desired crime types, changing the scale of the map, customising the details on the map and cross referencing with other types of incidents. The enhanced capabilities to visualise and analyse the spread and movement of crimes over a multitude of domains make the benefits of the Crime Mapping facility undeniable to the Safe City Programme and law enforcement.

 Analysis of Hot Spots (Source: FTCPD 2010)

 

Capability Improvement Using Machine Learning

On top of its existing capabilities, the Crime Mapping facility can be boosted with the additional research to add the capability to learn crime patterns and subsequently able to anticipate or predict the movement of crime. Of particular interest in this study is the ability to predict the movement of serial criminal suspects.

The collaboration between the Federal Town and Country Planning Department and the Royal Malaysia Police is the key success factor to the SCMS. When the need to predict arises, machine learning is invoked. Sets of database containing Mapping Information as well as Temporal Crime Information are interrogated.

Mapping Information is the database where spatial attributes pertaining to the previous locations of crime is stored while Temporal Crime Information is the database which stores information on crime. The relationship between temporal crime and spatial attributes is then figured out by machine learning which in turn predicts the next location and probably the time which the serial criminal suspect in question is expected to turn up. Law enforcement actions can then be taken by Federal Town and Country Planning Department or Royal Malaysia Police at the predicted location and time. Figure below visualises the conceptual framework as articulated.

 Conceptual Framework of GIS and Machine Learning Integration

 

NKRA Safe City Programme Achievements

Overall to date (Performance Management & Delivery Unit 2012), the GTP NKRA Crime Reduction programme has achieved the following:

  • Installation of 500+ CCTVs.
  • Mobilisation of 14,222 police personnel into 50 hot spots.
  • Deployment of 4,000 RELA & JPAM members into hotspots.
  • Ranked & awarded 753 Police Stations.
  • Redeployment of 6,751 police officers.
  • Activation of 358,811 Rakan Cop Members.
  • Reduction of 11% overall Index Crime from 2009.
  • Reduction of 40.6% overall Street Crime from 2009 baseline.
  • Reduction of 3.1% ”fear of becoming a victim of crime’: TNS Survey “Fear of Crime” Index: 48.9% (March to May, 2011 survey).
  • Accession of arrest cases brought to trial: 23.6% of Investigation Papers (IPs) opened are brought to trial.
  • Accession of public satisfaction with PDRM: Survey done in May 2011 shows 55% of respondents are satisfied with PDRM service from 35.8% in Dec 2009.
  • MyDistress Service launched in Selangor with 61,441 registered users as of Feb 2012.

Malaysia is validated as the 4th Safest in the Asia Pacific Region by Global Peace Index (GPI) for the first time ever after New Zealand, Japan and Australia on top of being the most peaceful country in Southeast Asia (Performance Management & Delivery Unit, 2012). To top this up, Malaysia is ranked 19th safest nation among the upper middle income countries by the World Justice Project’s Rule of Law Index.

The following business cases are solved (ESRI, 2012):

  • Real-time mapping and updating of crimes at all police stations through the integration with Police Reporting System.
  • Real-time geo-processing analysis.
  • Real-time Hot-Spot and Timeline analysis.

The Safe City Monitoring System was also recognised at Esri International Special Achievement Award in 2012.

FTCPD receiving Esri’s Special Achievement in Geography Award 2012

 

The author acknowledges the support of NKRA Safe City Program, Royal Malaysia Police and Federal Town & Country Planning Department Peninsular Malaysia to enable this concept be developed further into doctoral studies.