As supply chains grow increasingly complex and data-driven, it can be difficult to sort through the barrage of information to identify the best action to take or decision to make. But what if you had software that could learn to recognize patterns and make suggestions based on past experiences?
Dr. Noel P. Greis, director of the Kenan Institute's Center for Logistics and Digital Strategy at the University of North Carolina (UNC) at Chapel Hill, has been working to develop business intelligence software that can do just that.
Greis, who has a background in mathematics and engineering, says her interest in analytical software dates back to her work in graduate school on systems theory and design. Today she continues that work as she researches the use of business intelligence engines in the supply chain. In particular, she has focused on the use of "experience-based" analytics to improve decision making.
In addition to her work in business intelligence, Greis is the co-director of the UNC-Tsinghua Center for Logistics and Enterprise Development in Beijing, a joint venture of Tsinghua University's Department of Industrial Engineering and the Kenan-Flagler Business School at UNC. She was also the co-founder of the Global Logistics Research Initiative (GLORI), a worldwide consortium of 10 universities that conducted collaborative research on intelligent technologies in logistics.
In a recent interview with Editor James Cooke, Greis discussed software developments that could shape supply chain practices in the future.
Name: Noel P. Greis
Title: Director, Center for Logistics and Digital Strategy (U.S.), and Co-Director, UNC-Tsinghua Research Center for Logistics and Enterprise Development (joint venture between University of North Carolina and Tsinghua University in China)
Organization: Kenan-Flagler Business School, University of North Carolina at Chapel Hill
Education: Bachelor of Arts in Mathematics, Brown University; Master of Arts in Engineering and Master of Science in Engineering, Princeton University; Doctorate in Civil Engineering, Princeton University
Work History: Assistant Professor of Operations, Technology, and Innovation Management, Kenan-Flagler Business School, University of North Carolina at Chapel Hill; member of technical staff, Bell Laboratories and Bell Communications Research
CSCMP Member: Since 2000
What is meant by the term "experience-based" analytics?
Humans learn by experience. Our brain captures these experiences as sets of associations—for example "stove" and "hot." When faced with a decision, we generally draw upon our past experiences to search for analogues with similarities to the current situation.
Experience-based analytics use a type of pattern-recognition technology to accomplish the same tasks as our brains. We utilize software to capture and represent past "experience" and to help us make decisions in similar situations.
How will experience-based analytics impact supply chain operations?
As supply chains have become more complex and data-rich, humans are encountering limits in the amount of information that they are able to process. Experience-based analytics are able to augment humans' ability to process information in data-intensive applications like the supply chain.
For example, procurement officers in large, multinational organizations and government agencies process thousands of orders daily. These organizations have accumulated large amounts of history about their suppliers and how well they perform. Using experience-based analytics, we can "match" the best supplier with an incoming order based on the company's accumulated experiences about which suppliers have performed well with similar types of orders in the past, as distinguished by size, lead times, and other factors.
One of your research projects involved a battlefield supply chain management solution for Boeing. Can you describe that solution and how it came about?
Mention Boeing and most people think of its commercial aircraft products—the 747 or 787. However, developing large-scale systems that provide logistics support to the U.S. military is a very large part of Boeing's business. As Napoleon learned during his 1812 march on Moscow, the complex logistics of supplying everything from fuel and food to spare parts and ammunition to the battlefield can make or break a war. Success depends on strategic forward positioning of critical assets.
In the mid 2000s, the U.S. government turned to emerging technologies to try to solve these complex supply problems for the Iraq War. For Boeing, we created a system that provides battlefield situational awareness for logistics command-and-control. The key was real-time "sensing" of the operational status of in-theater vehicles and other assets and the fusion of that data with other contextual data. Our analytics "built" resupply missions that assured that the right amount of assets reached forward positions when and where they were needed, as safely as possible. The system was able to initiate resupply missions autonomously or semiautonomously. These experience-based analytics incorporated a technology called associative memory that was developed by Saffron Technology, one of our technology partners. Associative memory is a type of machine learning that captures the relationships between past experiences and present situations.
What can we expect of business intelligence software for supply chain management in the next two years?
A new information-rich environment and smarter analytics are changing the calculus of business decision making. Being able to take control of and respond to changes in the supply chain, especially disruptive events, requires more than visibility. We are starting to take advantage of tools that are better able to respond more quickly and effectively in dynamic environments.
For example, the Internet has matured as a connective technology, bringing with it an explosion of data. Data velocity is increasing and data types are proliferating. The virtual integration of the extended global enterprise is possible, and the availability and low cost of powerful multiprocessor computers and algorithms provide the hardware and software necessary to manipulate large volumes of data in near real time. Cloud computing allows companies to access services and data in real time via the Internet. And new software tools—the "experienced-based analytics" we've been discussing—are being developed that can learn and even make autonomous or semiautonomous decisions. This capability is still several years in the future, but we are building prototypes in our lab right now.
How likely will it be that business intelligence software will be able to predict "supply chain problems" before they occur?
Very likely. We are currently building software tools that are able to anticipate problems in the future. Managing a global supply chain is a complex sequencing act. At each stage of the supply chain, inventories must be kept supplied and in balance. Unlike traditional modeling approaches, we view supply chain coordination as a pattern-recognition problem. At any point in time, the supply chain can be represented by a large set of diverse and disparate data of "experiences," including inventory levels at suppliers, manufacturers, distribution centers, warehouses, and retail outlets; expected customer demands; and other factors that influence demand, such as promotions. Our analytics observe the supply chain over time and "learn" its dynamic behavior as a set of patterns. The tool's learned experience enables us to recognize situations that anticipate stock-outs or other supply chain failures.
Will business intelligence technology revolutionize supply chain practices?
Our appetite for business intelligence tools that help make sense of large volumes of business data will continue to grow. This is especially true because the costs of data acquisition and storage will continue to decline. And although the timeline is uncertain, business intelligence technology can be expected to enable great strides in supply chain practices. Right now we hear a lot about the "Internet of Things," where everything and everyone will be connected and able to communicate through a network enabled by the Internet—in effect merging the cyber and physical worlds. The Internet of Things is not just one technology; rather, it's a portfolio of technologies. Among them, business intelligence technology is an important first step.