Fighting Robbery in Latin America Through Data Science
Every Innocent Life Being Saved Is A Gain
Robbery is a real concern and a crucial security menace for a vast majority of developing regions. Latin American countries have presented the higher rates for this specific type of crime during the past decades (United Nations Human Settlements Programme, 2007). The generalized insecurity perception is alarming, and it is associated with the high degree of violence and victimization brought mainly by robbery and theft (Ojea, 2014).
Just in Colombia, almost 12,500 persons were murdered in 2018 during a robbery (Concejo de Bogotá D.C., 2019). Hence, enhancing security has become a priority for governments as well as a matter of personal interest to individuals.
In recent decades, developed nations’ police departments have started to address security utilizing Data Science. For example, London Authority’s Strategic Crime Analysis team has been able to identify hotspots of crime, to improve the police response and to analyse reports’ data to better understand incidents by cross-checking information from the Police Department, the Fire Brigade, the Health Care System, the transportation agencies and the ambulance services (Wooden, n.d.). Unfortunately, this does not apply to Latin America’s police departments for which countless opportunities still exist to innovate in crime-fighting as their techniques continue to be rudimentary.
Since political and social conditions are significantly worse in this region, challenges include not only data-related issues, but also bureaucracy and limited skilled human capital.
Difficulties notwithstanding, the impact of fighting robbery is immeasurable as it is translated into thousands of lives safeguarded annually.
Challenges Behind Fighting Robbery
Pion-Berlin (2010) affirms that there is an implicit security contract between citizens and governments. That is, national police departments are the authorities responsible for ensuring the safeguarding of the inhabitants of a country. Regardless of this being recognised theoretically in Latin America, the reality is that police forces ineffectiveness pairs with lack of funding and crew (Cao & Solomon Zhao, 2005).
Thus, challenges behind fighting robbery might include addressing the lack of effective presence of the police in hotspots, enhancing the understanding of criminal behaviour, augmenting the capabilities of teams in charge of security cameras monitoring, and speeding the recognition of street patterns.
Latin America, in particular, has little advancements on quantitative-orientated crime research as well as poor progress on criminology (Cao & Solomon Zhao, 2005). Given a widespread of all kind of crimes from on-street delinquencies to intra-family abuse (Moser & McIlwaine, 2006), exploitation of new technologies and analytical techniques to amplify police departments’ capacity becomes more pressing.
Data-driven Opportunities and Impact
Several targets arise for police departments in Latin America to innovate when fighting robbery. From people at risk to materialized events, three opportunities can be advocated:
A first approach to fighting any crime should be always prevention. Robbery is an ever-evolving phenomenon; changes in strategies and targeted zones by criminals occur daily. Therefore, automated data processing of reports is essential to reinforce police presence and to update citizens about high-risk locations. In this matter, hotspots could be identified by processing community and/or crime reports through clustering algorithms. Parameter creation and setting, based on final consequences -i.e. theft, homicide-, could be used to bring more attention to critical street blocks. Once dangerous zones were recognized based on citizens’ coverage, simulations and network analysis on streets configuration could be added in search of similarities along with the examination of correlation to crime-related threats (Leeds Institute for Data Analytics, 2019).
On this subject, it is crucial to emphasize the significance of local developments. De Melo (2015), after completing research work on spatial crime patterns in Brazil, highlights the difficulties to generalize findings in developed countries such as the U.S.A. and Canada to the Latin American milieu. Hence, the importance of devising data science solutions by each national police department. On that premise, a primary warning system could be built by merging these two complementary schemes: clustering and network analysis.
Altogether, providing citizens with updated, accessible information regarding real crime hotspots would forewarn them of danger on time.
A second opportunity lies under criminal reports analysis. A plentiful data pool could be constructed from merging clusters with information from national authorities. Natural Language Processing in conjunction with Machine Learning techniques might contribute to a better understanding of crime patterns and a rich and prompt identifying of incipient criminal conducts (Leeds Institute for Data Analytics, 2019). Success in cities like London demonstrates that a deep understanding of crime is the perfect match to a holistic picture of the scene, contributing to the updating and coordination of police strategies (Wooden, n.d.).
Behavioural human patterns recognition should be addressed when considering more complex approaches. In Colombia, 22% of 2018 total homicides were committed with bladed weapons and poniards (Policía Nacional de Colombia, 2019). Video processing to recognise movements related to the stabbing and/or weapons would be an unprecedented and astonishing application of Machine Learning and Deep Learning algorithms to fight on-street violence. The latter connected to security cameras would not only facilitate the work of policemen, but also could be set to automate sound alarms on the locations. Sirens have proven to generate early flight. If they are activated on time, delinquents could be dissuaded over crime perpetration (Lee, 2008).
In the third place, as it is not possible to achieve perfect solutions on prevention, it is imperative to consider opportunities to help the victims when a robbery is committed. In this regard, further linkage of facial recognition systems to security cameras might be useful resources in fugitive search. Impunity is the hidden face of injustice and crime in Latin America. As reported by the UN Office on Drugs and Crime in its latest Global Study on Homicide (2019), most homicide cases in various Latin American countries could not even be classified -82% in Colombia, 59% in the Dominican Republic, 69% in El Salvador-.
In cases of homicides or severe injuries, being able to close criminal processes is a must to protect victims’ and/or their family’s mental health.
Implementation Challenges
A challenging requirement to ensure warning systems genuinely impact is effective communication. To this end, police departments could employ software technologies -e.g. mobile phone applications- to make potential hotspots knowledgeable to the citizenship. Companies such as Rappi have implemented real-time heatmaps as an effective way to inform to collaborators the current state of the operation (Rappi, 2019). Even apps developed by amateurs have shown similar results on transmission. ResistenciApp was developed during the 2019 national strike in Colombia as a collaborative ecosystem to identify blockades and confrontations between police and protesters. Despite being a non-professional version, it was accepted rapidly as a great option to stay informed when routing during the strike (Martin López, 2019).
Thus, a smartphone application might be an aimed IT facilitator to communicate robbery-related warnings, with the advantage of allowing high-impact and easy-to-understand interactive visualisations.
Furthermore, implementation through mobile phone applications brings new opportunities to innovate. By integrating hotspots outcomes with user’s live location, back-end data science models and algorithms could be devised to trigger push notifications alerts related to closeness to high-risk areas. An analogous illustration of this opportunity is Uber’s feed ecosystem, which matches present data with existing information to organise feed’s flow, create categories and target real-time pushes (Forsythe et al., 2017).
Thus, smartphone apps would not only be a route to implementation, but also a catalyst for further impact. Live hotspots maps coupled with alarms/notifications/SMS sent to citizens who are getting close to a threat would reinforce prevention for those zones lacking presence of patrolmen.
On the other hand, dynamic data collection and validation arise as an extra challenge for implementation. In this regard, without discarding other possibilities, mobile phone applications might offer an extra benefit to the above-mentioned gains. Community-based crowdsourcing applications have proven to be an innovative approach to data collection techniques. Rothstein, Jennings, Moorthy, Yang and Gee (2016) assessed the feasibility and acceptability of a community-based health care app concluding that the acceptance rate was positively correlated to the degree of utility perceived by users.
Accordingly, given that security is a matter of personal interest, the volume of data collected collaboratively will likely increase if conditions such as privacy and engagement are favourable.
Correspondingly, reliability could be reinforced by the community. Waze dual evaluation of alerts can exemplify this advantage. When receiving input from the public, Waze marks each alert with two scores. The first is based on the number of confirmations from other users, and the second is based on the experience level of the reporter (Google, 2020). By these means, poor-quality information is filtered before reaching other users. Similarly, citizens’ reports on robbery spots could be validated by various citizens and/or verified by the police department. This whole ecosystem would ensure good-quality updated data and would facilitate scalability.
Critical Potential Data Issues
The Latin American context brings particular social and political difficulties in the discussion. To start with, the labour force is deficient for skilled positions. McKinsey Global Institute (2018) reported that the difficulties to find trained workers in Latin America surpassed the global average, as cited by 40–50% of employers when inquired about vacancies. This limitation becomes more critical for those challenges that require advanced technical knowledge such as recognition engines and behavioural analytics. A prospect to face this obstacle in the short-term is to leverage human capital from developed countries. When funding limitations also occur, organisations such as DataKind (2020) facilitate the access to Data Science professional teams by offering pro-bono long-term engagements.
However, deeper solutions should entail increasing not only the analytics capability, but also exploiting could computing capacities, to process and store big data boosting long-term competitiveness (Manyika et al., 2011).
To continue with, worse case scenarios should not be discarded. Acknowledging that bureaucracy issues in Latin American countries might completely prevent any (or all) of the above-mentioned opportunities from being implemented, there is always the option to rely on private or not-for-profit organisations. A community-based crowdsourcing application would enable this direction given that its foundations are not subject to state intervention. The UN Refugee Agency (2008, pp. 14–15) outlines community-based approaches as those created “to empower all the actors to work together to support the different members of the community in exercising and enjoying their human rights”.
Therefore, by appealing to community active collaboration the development of an effective initial approach for prevention would not depend on government.
Last and equally important, privacy of data must be a constant concern. Regarding information collected through community-based collaborative means, privacy should be ensured via anonymisation. Coherently, smartphones apps permissions should not access other personal data beyond location, which should only be temporarily employed to trigger alerts. Other technologies such as facial recognition prompt even more issues. While addressing the challenges of these technologies, Carter (2018) recognizes civilians fears in respect of the misuse of personal information by authorities.
The trade-off between privacy and safety is a constant public debate.
Hence, when police departments and justice system are considered, data protection turns even more critical given the high degree of sensitivity. End to end security and encryption, and protection against hackers should be deemed appropriate by authorities in charge.
Every Innocent Life Being Saved Is A Gain
The OECD Better Life Index (2020) reveals that the ratio of the national population not feeling safe while walking unaccompanied at night signifies 56% in Colombia, 64% in Brazil, 52% in Chile and 58% in Mexico. By comparison, all are significantly higher than the OECD average of 32%.
The real impact of all innovations mentioned above could not be quantified. In Latin American countries, potential victims are almost equivalent to the whole population, especially in big cities.
Beyond the decreasing of robbery rates, which is an objective per se of police departments, any reduction of the number of innocents being murdered is a gain.
Make your story worth it, may your dreams be your hope.
If you change your world, you change the world.
References
Cao, L., & Solomon Zhao, J. (2005). Confidence in the police in Latin America. Journal of Criminal Justice, 33(5), 403–412. https://doi.org/10.1016/j.jcrimjus.2005.06.009
Carter, A. (2018). Facing reality: Benefits and challenges of facial recognition for the NYPD. Homeland Security Affairs, 5–24. http://ezproxy.lib.uts.edu.au/login?url=https://search-proquest-com.ezproxy.lib.uts.edu.au/docview/2206258314?accountid=17095
Concejo de Bogotá D.C. (2019, January 23). Altas cifras de robos en Bogota. https://concejodebogota.gov.co/altas-cifras-de-robos-en-bogota/cbogota/2019-01-23/084055.php
DataKind. (2020). DataCorps. https://www.datakind.org/datacorps
De Melo, S. N., Matias, L. F., & Andresen, M. A. (2015). Crime concentrations and similarities in spatial crime patterns in a Brazilian context. Applied Geography, 62, 314–324. https://doi.org/10.1016/j.apgeog.2015.05.012
Forsythe, A., Haenikel, D., & Zha, M. (2017, May 2). Redesigning Uber engineering’s mobile content delivery ecosystem. Uber Engineering Blog. https://eng.uber.com/mobile-content-delivery/
Google. (2020). About confidence and reliability scores. Google Help. https://support.google.com/waze/partners/answer/6324421?hl=en
Leeds Institute for Data Analytics. (2019, June 13). Using data science to fight crime. https://lida.leeds.ac.uk/research-projects/using-data-science-to-fight-crime/
Manyika, J., Chui, M., Brown, B., Bughin, J., Dobbs, R., Roxburgh, C., & Hung Byers, A. (2011). Big data: The next frontier for innovation, competition, and productivity. McKinsey Global Institute. https://www.mckinsey.com/~/media/McKinsey/Business%20Functions/McKinsey%20Digital/Our%20Insights/Big%20data%20The%20next%20frontier%20for%20innovation/MGI_big_data_full_report.ashx
Martin López, C. (2019, December 2). Crean app para identificar bloqueos durante protestas en Bogota. Caracol Radio. https://caracol.com.co/emisora/2019/12/02/bogota/1575316604_270381.html
McKinsey Global Institute. (2018). Preparing Brazil for the future of work: Jobs, technology, and skills. https://www.mckinsey.com/~/media/McKinsey/Featured%20Insights/Future%20of%20Organizations/Preparing%20Brazil%20for%20the%20future%20of%20work%20Jobs%20technology%20and%20skills/MGI-Future-of-Work-Brazil-Briefing-note.ashx
Moser, C. O., & McIlwaine, C. (2006). Latin American urban violence as a development concern: Towards a framework for violence reduction. World Development, 34(1), 89–112. https://doi.org/10.1016/j.worlddev.2005.07.012
OECD. (2020). How’s life? OECD Better Life Index. http://www.oecdbetterlifeindex.org
Ojea, V. (2014, March 5). Latin America accounts for more than 30% of the world’s homicides. The World Bank. https://www.worldbank.org/en/news/feature/2014/02/11/en-america-latina-sufre-mas-del-30-de-los-homicidios-mundiales
Pion-Berlin, D. (2010). Neither military nor police: Facing heterodox security challengers and filling the security gap in democratic Latin America. Democracy and Security, 6(2), 109–127. https://doi.org/10.1080/17419161003715710
Policía Nacional de Colombia. (2019, April 25). Estadística delictiva. https://www.policia.gov.co/grupo-informaci%C3%B3n-criminalidad/estadistica-delictiva
Rappi. (2019, September 4). Aumenta tus ganancias con el nuevo mapa de pedidos. Soy Rappi Blog. https://blog.soyrappi.com/aumenta-tus-ganancias-con-el-nuevo-mapa-de-pedidos/
Rothstein, J. D., Jennings, L., Moorthy, A., Yang, F., Gee, L., Romano, K., Hutchful, D., Labrique, A. B., & LeFevre, A. E. (2016). Qualitative assessment of the feasibility, usability, and acceptability of a mobile client data app for community-based maternal, neonatal, and child care in rural Ghana. International Journal of Telemedicine and Applications, 2016, 1–14. https://doi.org/10.1155/2016/2515420
United Nations High Commissioner for Refugees. (2008). A Community-based Approach in UNHCR Operations. https://www.refworld.org/pdfid/47da54722.pdf
United Nations Human Settlements Programme (UN-Habitat). (2007). Enhancing Urban Safety and Security: Global Report on Human Settlements 2007. https://books.google.com.au/books?id=SmsbwAtSfE0C&pg=PA56&lpg=PA56&dq=ROBBERY+CONCERN+LATIN+AMERICA&source=bl&ots=OjnB3H1pXx&sig=ACfU3U2f4QKe7g5HhQ060PSIWWG1cjnEZQ&hl=es-419&sa=X&ved=2ahUKEwiXpaDQpuXoAhXtxjgGHfxuAE8Q6AEwEHoECA0QLw#v=onepage&q=ROBBERY%20CONCERN%20LATIN%20AMERICA&f=false
United Nations Office on Drugs and Crime. (2019). Global Study on Homicide. Understanding homicide: typologies, demographic factors, mechanisms and contributors. https://www.unodc.org/documents/data-and-analysis/gsh/Booklet_3.pdf
Wooden, A. (n.d.). New Technology is changing how cities tackle crime. Intel. https://www.intel.co.uk/content/www/uk/en/it-management/cloud-analytic-hub/big-data-analytics-crime-fighting.html
