Skill Set

Analysts are deluged with information on a daily basis. The ability to bring some order into this informational chaos can have a huge impact on public safety and the quality of life in the communities that they serve. On the other hand, the opportunity to bring analytical and predictive models directly into the operational environment holds the promise of giving public safety and intelligence professionals the ability to maneuver within the decision and execution cycles of their opponent. Whether it is the war on terrorism, the war on drugs, or the war on crime, enhanced knowledge and the ability to anticipate future actions can afford operational personnel essential situational awareness.

Knowledge of advanced statistics is not a prerequisite for using predictive analytics. In fact, the discovery process associated with data mining also could be viewed as after-the-fact explanations for unpredicted outcomes, something somewhat distasteful in inferential statistics. When examined under the intense scrutiny of the analyst’s domain knowledge, however, these unanticipated or surprising findings can have significant value and greatly enhance our understanding of crime and intelligence data. For those who are analytically inclined, it can be a wonderful and exciting process of data exploration and discovery. Those with a strong background in statistics, though, might be somewhat handicapped by the comparatively rigid nature of inferential statistics, with all of its associated rules and assumptions. With a little confidence and practice, even statisticians will be able to overcome their previous training and perform what they once considered to be unnatural acts with data and information.

On the other hand, data mining brings powerful analytics to those who really need them, including operational personnel. In my experience, it is far easier to teach someone with interest who knows something about crime and criminals how to effectively use these tools. With some guidance regarding a few “rules of the road” for data mining, and the application of off-the-shelf software tools, data mining is well within the reach of any organization with an interest and willingness to put more science and less fiction into crime and intelligence analysis. Moreover, many of the new tools have been adapted to run in a web-based environment and are no more difficult than making a purchase or completing a survey over the Internet. These advancements have created the opportunity for “24/7” analytical capacity,ⁱⁱ even within smaller agencies with comparatively limited personnel resources.

The more that operational personnel, managers, and command staff understand the information requirements and possible outcomes from analytical products; the more likely they will be to contribute data that is meaningful, detailed, and valuable. They also will be in a better position to work with the analyst and participate in the analytical process, requesting output that has increased value for them as they acquire a better understanding of what is available. By understanding the importance of the data inputs and the potential range of outputs, operational personnel, managers, and command staff alike can become informed information consumers and increase the likelihood of identifying actionable output from the analytical process. This subtle change in relationships and understanding can greatly enhance analysts’ ability to gather the necessary data and information, ultimately increasing their ability to support operational personnel, policy decisions, managers, and command staff.

At a recent security expo, Tom Clancy advised the security and intelligence professionals in the audience to seek out the “smart people,” observing that, “[t]he best guys are the ones who can cross disciplines . . . [t]he smartest ones look at other fields and apply them to their own.”ⁱⁱⁱ In my opinion, many of the “smart people” Clancy refers to will rise out of the operational ranks, given the intuitive nature and relative ease of use associated with the new generation of data mining and predictive analytics software tools. While most analysts probably do not need to fear for their jobs just yet, increasingly friendly and intuitive computer systems will allow data and information to serve as a fluid interface between analytical and operational personnel. At some point in the future, that distinction will become almost meaningless with the emergence of increasingly powerful software tools and systems and the “agent/analysts” that employ them.

“Agent/Analysts” and Future Trends

I see a day in the not-too-distant future when analysis will be available without immediate access to an analyst. Information from operations will feed analysis, while the analysis will concomitantly drive the operations, thereby creating a feedback loop of ever-increasing information and actionable intelligence. I see a day when a patrol officer will come back to work after several days off and, at the beginning of the tour, will be able to review recent patterns and trends within the context of historical data and accumulated knowledge from the mobile data terminal in his cruiser. After responding to his first call, he will be able to enter the incident information directly into the department’s computerized records management system (RMS) using direct voice commands. This information then will be used to create the computerized offense report. Any digital images captured from the incident will be quickly uploaded and linked directly to the offense report, as well as any associated or linked information already stored in the RMS. During the data entry process, this new information will pass through an analytical filter prepared earlier in the week by the analytical staff, who are home asleep at this hour. The algorithm running in the background will quickly link this most recent incident to a recent series and prompt the patrol officer to consider several possible alternatives. With this real time, value-added analysis, the officer can make quick, information-based operational decisions that result in a rapid apprehension of the criminal.

This is handled similarly when an agent in a remote location is debriefing a suspected terrorist. The verbal information is recorded and transcribed directly into a free format text file using voice recognition software. The file is then uploaded to an analytical fusion center a thousand miles away. An analyst there uses sophisticated text mining technology to probe and characterize the results of the interview. Several key phrases are identified and compared to an existing database generated from earlier interviews with members of the same terrorist cell being held in other locations around the world. Based on the analysis of the current interview and its comparison to the existing models, areas of possible deception and truth are identified and highlighted, as are promising interviewing strategies. This information, including the interviewing strategies and approaches, is sent back to the agent in the field, further informing and guiding the ongoing interview process, while concomitantly enhancing the existing intelligence on the operations, practices, and strategies of this particular terrorist group.

Are these extravagant predictions? Absolutely not. Both scenarios outlined above are based on existing technologies and resources. In many ways, approaches and methodologies similar to information management have been used in the business community for years. All that is required to implement these strategies is a commitment to take advantage of the currently existing analytical tools and incorporate them into our world. Unfortunately, a paradigm shift in how we view information, analysis, and the relationship between analytical and operational personnel also will be required. That probably will be the most difficult task. Once we overcome that hurdle, however, adapting these new technologies promises to be one of the most exciting adventures in public safety in our lifetime.

How To Use This Book

All of the examples included in this book come from real experience. In some cases, though, the specifics have been changed to protect ongoing investigations, sensitive data, or methods. Whenever possible, I have tried to distinguish between real cases, particularly those taken from published work, and those generated specifically as examples. Given the nature of some topics covered in this book, however, it would be inappropriate to provide too much specific detail and compromise methods. To be sure, though, while the names might have been changed to protect the “not so innocent,” the examples are based on real experiences.

This book is divided into five main sections: “Introduction,” “Methods,” “Applications,” “Case Examples,” and “Advanced Concepts/ Future Trends.” The third and fourth sections include annotated examples focusing on the why and how, as well as the limitless possibilities for data mining and predictive analytics in crime and intelligence analysis. While this organization is relatively logical for training purposes, many readers will choose to read the book out of sequence. In particular, managers, command staff, supervisors, policy makers, and operational personnel interested in learning more about data mining and predictive analytics but not expecting to use these tools first hand will have neither an interest in nor a need for detailed information on specific methods and algorithms. These readers could benefit from reading and understanding the annotated examples if they make acquisition and purchasing decisions for analytical products and determine the focus of their analytical personnel. Moreover, operational personnel can more fully exploit the new technology and work more effectively with analytical personnel if they understand the vast array of possibilities available with these new tools. With the opportunity to deploy data mining and predictive analytics directly into the field, an increasing number of operational personnel will be using data mining products. While they might not be generating the specific algorithms or models, a general understanding of data mining and predictive analytics will certainly enhance their ability to exploit these new opportunities.

Similarly, many analysts will use this book to explore the possibilities for data mining in their environment; identifying ideas and strategies from the annotated examples in the third section, and then returning to the methods section for specific information regarding the use and implementation of these approaches. This book is not intended to provide detailed information about specific software packages or analytical tools, but merely provides an overview of them. It should serve as a starting point, using terminology, concepts, practical application of these concepts, and examples to highlight specific techniques and approaches in crime and intelligence analysis using data mining and predictive analytics, which each law enforcement or intelligence professional can tailor to their own unique situation and responsibilities. While the basic approaches will be similar, the available data, specific questions, and access to technology will differ for each analyst and agency, requiring unique solutions and strategies in almost every setting.

Perhaps one of the most challenging aspects of writing this book was keeping abreast of the new developments and data mining applications that now appear on an almost daily basis. It is both frustrating and exciting to consider how much this field is likely to change even in the short time between completion of the manuscript and actual publication of the text. Therefore, the final section, “Advanced Concepts/ Future Trends,” should not be viewed as inclusive. Rather, this particular section is intended to serve as a beginning for ascending to the next level of training for those interested in this field. This rapid pace of innovation, however, is what keeps the field of analysis fresh and exciting, particularly for those with the interest and creativity to define the cutting edge of this new and evolving field.