Opinion (Rogério COPETO / Official GNR): PREDICTIVE POLICING.
"Predictive Policing is the application of analytical techniques - particularly quantitative - to identify likely targets of police intervention and prevent crime or solve past crimes, making statistical predictions".
Rogério COPETO
Lieutenant Colonel of the GNR
Master in Law and Security and Homeland Security Auditor
Head of the Education Division / Command Doctrine and Training
The setting is the responsibility of RAND and it was cited by Colonel Pedro GNR Moleirinho, in the text of his own with the title "The importance of predictive models in safety. The risks and unstable equilibria"And published in the book"Predictive models & Public security", edited by the Border Docaos, dated October 2018.
The RAND document to which we referred earlier called "Predictive Policing – The Role of Crime Forecasting in Law Enforcement Operations"Is dated 2013 and serves as a reference to this Article.
According to the document are several Predictive Policing methods that are currently being used by security forces (FS), all over the world and much has been written on the subject, but we add, not always fair, on the effectiveness of the analytical techniques used, to identify potential criminal activities, preventing them before they happen.
But as mentioned by the RAND prediction and forecasting, are different concepts, where the first is objective, scientific and reproducible, while the second is subjective, mainly intuitive and unplayable, verifying that the FS adopted Predictive Policing term to describe this model of policing, probably be objective, scientific and reproducible, which in itself is revealing of their added value.
So, and with respect to predictive methods, they can be divided into four major categories: Methods prediction crimes: use prediction techniques of places with high probability of occurrence of crimes; offenders prediction methods: These approaches identify individuals with a high probability of committing crimes in the future; criminal identification prediction methods: Use techniques to create individual profiles, with criminal past, and; prediction methods of crime victims: Similar to the methods used for identifying offenders and locations with high probability of occurrence of crimes, these approaches being used to identify groups or, in some cases, individuals who probably may become victims of crime.
Densifying the four categories of methods used in Predictive Policing, indicates that for predicting the occurrence of crimes, Conventional approaches to the problem are used, including the identification of locations where crimes are concentrated, considered "hot spots", similar to what is done with the "accident accumulation of local", in the context of road safety.
The second category, gathering methods to identify individuals at high risk of committing crimes in the future, They are used conventional risk assessment techniques, particularly in the identification of risk factors associated with the individual in question, It is the highest risk, when they are more risk factors present, similar to what is currently being done with the risk assessment tool for domestic violence, by FS. These methods can therefore also be used to assess the risk of a given individual will become a victim of crime, well it is used in violent groups and how to predict the likelihood of violent brawls happen, no future, between so-called "gangs".
The third category of methods and used to identify potential criminals, rely mainly on available and information collected at crime scenes, with the aim of linking suspects to crimes under investigation, using the elimination process, using the database, stating, the name, criminal records and other information known suspects. Predictive analytics is the combination of the available evidence to potential suspects (not previously identified), can be analyzed large amounts of data.
In the third category and the identification of groups and, in some cases, of individuals, that are likely to become victims of crime, They are used the same methods to predict where and when crimes may occur, as well as some of the methods used to predict who is more likely to commit crimes, resorting to risk appraisal conducted. For the prediction of victims of crime, It is necessary to identify groups of individuals at risk and, such as the groups associated with certain types of crime such as domestic violence, that are close to risk sites or who are at risk of victimization.
All of these methods rely on mathematical models of predictive analytics, commonly called "algorithms" artificial intelligence, analyzing large amounts of data, a “big data”, who might be from the database of FS or social networks, according to critics of Predictive Policing, this being one of the first risks associated with this model of policing, but this is subject for another article.
The RAND states that make predictions or forecasts is only half the Predictive Policing, because the other half involves actions on the ground, in compliance with forecasts, in order to reduce crime or at least solve the crime that was planned, This policing model being composed of four steps, as quoted by Colonel Pedro GNR Moleirinho:
– "The first two are the collection and analysis of data fusion for the crimes, the incidents and offenders, so that they can formulate forecasts.
– The third step is to conduct police operations that have an impact on crime provided (or to help solve the crimes of the past).
Information should be made available to meet the requirements of a situational awareness of the risks to various operational levels, to act on them and anticipate crime, this being an essential part of any intervention plan.
– In the last step, interventions lead to a criminal response that ideally reduces or eradicates crime".
It should be noted that in the third step the type of intervention or police operation, will vary according to the situation and the valence who was given the responsibility to intervene. This intervention can be of three types, ranging from the simplest to the most complex: general intervention; specific intervention, and; Intervention directed to the problem.
Eventually the most complex interventions require more resources, should, however, be the most appropriate to the expected problem, in order to achieve the best results. Whatever type of intervention, those who perform need information the most accurate possible, in order to perform the operation successfully.
So, provide current and accurate information to the elements that run on the ground the crime prevention actions, It is a critical part of the Policing Predictive.
And at the end of the fourth stage it will be necessary to the mandatory reviews, to ensure that interventions are being carried out correctly and that it does not result from any problems, repeating the cycle.
To finish, it is concluded that the FS will benefit from the implementation of Predictive Policing, should, however, use tools with the ability to provide accurate and current information about the probability of crimes, to anticipate it or solve it, verifying that these tools are available free or at low cost and include Microsoft Office features or basic tools of geographic information, because even the information technology systems and sophisticated communication, They can only provide reliable data, if the data from sources they also reliable and without having been intoxicated by prejudices transformed into computer language.
