The two topics Predictive Analytics and Big Data often appear together in discussions and blogs. This is not surprising when one considers that one of the great differences between traditional data warehousing and the new big data era is the outcome that each one has been tasked to produce. Databases and data warehousing have traditionally been used to provide a report (basically, descriptive analytics) about what has happened or what is happening. Conversely, when big data captures sufficiently complete information about a domain and its participants so that we can (and do) use the data collection to predict what will happen (and maybe when and where it will happen) – this is predictive analytics.
Predictive analytics is applicable in many domains where diagnosis and decision support are critical, including science, retail, social media, cybersecurity, health, and law enforcement. As important as it is, predictive analytics is but one component of a powerful set of data science methods that are being applied to big data. We have previously described those components within the context of the corresponding maturation and growth across the big data analytics domain: descriptive to predictive to prescriptive to cognitive analytics.
We capture here our interview with Colleen McCue, one the great pioneers in the field of predictive analytics, particularly in the area of public safety and law enforcement (e.g., predictive policing). Part 1 of our interview appears in this article, and Part 2 will appear in our next article.
We asked Colleen McCue to describe her work (past, present, and looking forward) in the field. I first met Colleen via a conversation over five years ago when she was writing the article “Predictive Policing: What Can We Learn from Wal-Mart and Amazon about Fighting Crime in a Recession?” for Police Chief Magazine. She wanted to quote some of my work in her article and contacted me to discuss it. At that time, I was very surprised to learn that my work (primarily in scientific data mining up until then) had any impact outside of the physical sciences (where I worked for my entire career). In fact, I was so impressed with myself, following my conversation with Colleen and the publication of her article, that I subsequently referenced her article in many of my own reports, including that very special research report that I wrote for my successful promotion to Full Professor at George Mason University. So, I thank Colleen for her insights and making that connection. And I thank her for taking the time to tell us more about herself, predictive analytics, and predictive policing.
1) Colleen, please tell us about yourself: What is your background? How long have you been involved in data analytics? What brought into this field of work? What has been the most exciting aspect of your work over the past several years?
I am a physiological psychologist by training and have been studying violent crime and other bad behavior for almost twenty years at this point. Although I came out of an academic environment, I really appreciate the urgency associated with supporting the operator, and the associated satisfaction in seeing the results of analysis being used to change public safety and security outcomes.
2) Colleen, do you call yourself a data scientist? Why, or why not?
Despite having a number of different titles over the years, I always have considered myself to be a scientist. That is how I was trained, and how I still function today. I really like the concept of data science, particularly as it relates to creativity, innovation and novel transdisciplinary solutions, but am neutral on the nomenclature. At my heart, though, I am and always will be a scientist; data scientist, behavioral scientist, or just plain scientist.
3) Colleen, you wrote a book on "Data Mining and Predictive Analysis" a few years ago. Tell us about it – what motivated you to write the book? Where have you seen your book being used or referenced?
I wrote the book because I knew that the operational public safety community could benefit from advanced analytics. My husband was deployed overseas at the time and I was a functional single mother of five when I wrote it. I was committed, though, to bringing more science and less fiction to crime and intelligence analysis, and saw real promise in the role that operational security analytics could play in changing public safety and national security outcomes.
It is still really exciting to see my book being used or referenced, and it is encouraging to see that it is being used domestically, as well as abroad as the core course text for crime and intelligence analysis training.
I am in the process of writing the second edition and it is absolutely amazing to mark progress in our field by how the specific capabilities, sources, and analytic tradecraft have developed since the first edition. In particular, geospatial capabilities have become concomitantly more powerful and also more accessible and easy to use.
4) Colleen, your book predates nearly all current books and products today related to predictive analytics. Tell us how things have changed or stayed the same in the field of predictive analysis since you wrote your book.
It has been interesting to watch the field develop, and this is a very exciting time to be working in predictive analytics. New approaches to visualization are making these powerful capabilities directly accessible to the operator, supporting real time decisions. I am particularly excited about the move in favor of a “dynamic” and “immersive” analytic experience for the operational end user, as envisioned by National Geospatial-Intelligence Agency (NGA) Director Letitia Long.
On the other hand, this enthusiasm also has created a veritable flood of junk science. One of my goals in writing the textbook originally was to create informed consumers, particularly in the management and command staff roles. I have no illusions that someone will be able to read my book and become a highly trained analyst. On the other hand, if the work that I have done opens their eyes to the “art of the possible” and/or enables them to make a better choice regarding the acquisition of talent or technology, then I have achieved my goal.
This ends Part 1. Part 2 of our interview with predictive analytics pioneer Colleen McCue will continue in our next article. We are grateful to Colleen for taking the time to answer these questions and to provide us with her insights into the ever-growing field of predictive analytics.