At Veracity we go well beyond the commonly understood definition of Data Science. The term itself is relatively new and includes a number of disciplines such as statistics, data mining, and predictive analytics.
Yes, we develop algorithms and write code to fit models, but at the root of it we are in the business of trying to figure out “how the world works.” We seek to illuminate hard-to-see connections––between what was planned in a given enterprise and what actually occurred. The difference can be huge, and it may never have been understood before. We then connect this knowledge with computational forecasting models that clarify decision paths for resource managers. These paths are true maps of how a particular world began, where it went from there, and where it is now most likely heading.
Using our work in naval aviation as an example, aircraft are flown on missions and burn up resources every time they fly. We want to know how to predict the supply of these resources for best planning and budgeting in the future. To do this we study actual, historical data (e.g., flight hours logged, spare parts consumed, maintenance accomplished) and connect this information to the demands that drove the use of resources––complex issues such as operational planning, budgets, and even mishaps. From these connections we are able to build predictive, mathematical models, based on probabilities, for the demand and availability of resources in the future.
My own path in the field of computational forecasting is unlike the typical Data Science background. In fact, I never took a course in Data Science (the term didn’t exist then) or even statistics. I studied atomic physics in research labs and simulated the properties of materials under extreme conditions, using high-performance computing. While that wasn’t Data Science per se, the experience gave me an excellent perspective on how to think algorithmically. In other words, I learned how to view the world in terms of inputs, followed by all of the critical steps and decision points that lead to outputs. Sounds pretty basic, but of course, trying to model the ever-complicated real world in a way that leads to meaningful decisions requires a lot of thought, hard work, and real science, the part I enjoy the most.
As scientists, we guess about how we think the world may work, then consider what we would see in the data if that were or were not the case, and then we see if we were correct and adjust as required. I enjoy that at Veracity we work in a continual learning and development environment. We are always striving to better understand and represent how the world works so we can better forecast how it may unfold. Call it Data Science, plain old science, or high-performance computing, I am absorbed by the power and utility of computational forecasting.
“I am absorbed by the power and utility of computational forecasting.”
Dan Horner is the Data Science Lead for the Veracity Forecasting Group. He holds a Ph.D. in Theoretical Chemistry from the University of California, Berkeley, and a B.A. in Physics (summa cum laude) from Boston University. Before joining Veracity, he served as a Research Analyst and Research Scientist at the Center for Naval Analyses from 2010-2013 and held Postdoctoral fellowships at the Lawrence Berkeley National Laboratory (2004-2006) as well as the Los Alamos National Laboratory (2006-2009).