In a nutshell, predictive policing is the practice of using data to both predict and forecast future activity so that agencies can deploy resources in the most effective manner to ultimately prevent and deter crime from occurring. There have been countless publications and research articles written in recent years documenting the success of this policing strategy. There is no doubt that the use of data and analysis to help guide the decision-making process in law enforcement is here to stay. However, there are still many police departments throughout the country that are not leveraging these techniques. While there are many reasons for this reluctance, one of the most common notions is that policing professionals feel that their agency is too small to need or benefit from predictive policing techniques. Nothing could be further from the truth.
No matter the size of the agency, the practice of using data to make better informed tactical, strategic and operational decisions is a sound philosophy. While predictive policing is relatively new, the basic principles of this philosophy have been around for a very long time in the business world. Businesses decide how much product to order at what time of year, how many staff members need to be employed and what projected sales are going to be next quarter all through the analysis of data. A law enforcement agency claiming that predictive analytics is only for the “big dog” agencies is akin to a family-owned grocery store ordering products at random and ignoring their sales history and other data because they are not as big as Walmart or Target. That way of thinking just doesn’t make sense.
Another common reason articulated by agencies who feel that predictive policing is “not for them” is that the analytical software and skilled crime analysts are too expensive. Almost every municipality in the country is being asked to do more with less and have been dealing with shrinking budgets for years. This has affected agencies both large and small. In fact, it could be argued that a reduction in sworn officers could more greatly impact a smaller agency that did not have the “extra” resources in various specialty units throughout their agency. While spending money on analytical software and/or a crime analyst may seem counterintuitive when budgets are tight, these actions can actually increase the effectiveness of an agency’s limited resources.
Putting a predictive policing strategy into practice starts with obtaining the right technical software and trained personnel to utilize it. There are several key factors that must be considered when selecting analytical software, including upfront costs, ongoing costs and ease of use, but none is more important than feeling comfortable with the algorithm being used and feeling confident about the results that are being produced. While budgets may be frozen and certain resources are dwindling throughout the law enforcement community, data is abundant in this industry. Every agency, no matter how large or small, has data available. Case reports and calls for service are being documented somewhere and that data is typically available. An agency should be diligent in testing the predictive analytics software that they have against their own data to make determinations about how and when to utilize it.
There is no software available that can predict every crime type, for every time period, in every part of the county, every single time. There is no magical equation that can be used in every situation. However, through testing, an agency can determine what works best for their jurisdiction and utilize their software tools to the best of their potential. If a grocery store used an analytical model to determine how much eggnog to purchase around Christmas and then ended up having case after case still on their shelves after the holidays that they couldn’t sell, they most likely would use a different approach next year to avoid wasting merchandise. The same approach should be used by law enforcement. If an agency wants to combat residential burglaries and it uses software to produce a forecast anticipating where they are likely to be concentrated next month, then the department should have already tested that model and predicted previous months that have already come and gone so that the predictions could be compared to the actual concentration of crime. By doing this, the agency will develop a confidence level and know when it is and is not appropriate to use certain predictive algorithms.
Purchasing the right software and fully testing it is only part of the process of implementing a predictive policing strategy. The much more difficult job is creating institutional knowledge about the philosophy and getting buy-in from all levels of the agency. Predictive policing cannot work if the chief of police and a few members of the command staff are the only ones on board. Command staff has to authorize overtime and change shift allocations when necessary, line level supervisors need to direct proactive policing efforts within predicted zones during predicted time periods, and individual officers need to understand the importance of their role once in these predicted areas. No matter how accurate a forecast or prediction may be, it will not help an agency achieve its goal of deterring crime if it is not acted upon quickly.
It is clear that predictive policing is a valuable strategic approach to fighting crime for any law enforcement agency, regardless of size. That’s not to say, however, that there will not be some differences in how predictive policing is used in a smaller agency. In a smaller agency, next-event predictions might not be needed for bank robbery or sexual assault series every week. But the truth of the matter is that the same exact analytical practices can be applied to any crime series with the same results. In other words, smaller agencies that recognize a series of connected, albeit, less serious crimes, such as a vandalism or bike theft series, could perform next-event prediction analysis to determine the best place to deploy resources to disrupt the activity and apprehend the offender. The same idea holds true for more general forecasts. While a larger agency might be forecasting where the 30 or 40 gang-related violent crimes are likely to be next week, a smaller agency might be forecasting where the five or 10 property crimes are going to be next month. The seriousness and the frequency of the crime is less in a smaller agency, but that does not mean that the value of using data analysis to guide police intervention strategies is any less important.
Fortune 500 companies have a responsibility to their shareholders to be as profitable as possible. It is common knowledge in the corporate world that analyzing data to turn raw information into business intelligence is one of the necessary practices in order to fulfill that responsibility. The citizens who live and work in the communities that your agency serves are the shareholders. While law enforcement agencies have a far more awesome responsibility than quarterly profits and dividends, the field could take a lesson from the corporate world and accept that examining data and performing predictive analytics is simply the best way of doing business.