While companies see the value in using predictive analytics and big data in their supply chains, the cost of deployment is still deemed too high. New research suggests ways to understand — and use — data better.
While some industries like health care and retail are starting to see the transformational potential of big data and predictive analytics, these strategies haven’t quite panned out for supply-chain managers. Why?
The biggest obstacles appear to be the cost of hiring skilled employees and the complexity of connecting nodes across an extended supply-chain network.
With this in mind, researchers Matthew Waller, chair of the department of supply chain management at the University of Arkansas’ Sam Walton College of Business, and Stanley Fawcett, John B. Goddard Endowed Chair in global supply chain management at Weber State University, write in a new research paper that the convergence of data science, predictive analytics and big data have the potential to transform the way in which supply-chains managers lead and supply chains operate.
But, they say, more research needs to be done to investigate the convergence of data science, predictive analytics (they’re calling this convergence DPB) and big data in the field of supply chain management (SCM). The goal: to further the understanding of how to utilize big data effectively and develop a new breed of supply chain leaders that are versed in using data and analytics effectively.
Their research has implications for current managers.
A recent Wall Street Journal article citing a survey by The Economist points out that while most companies see the value in using predictive analytics and big data to parse out increasingly complex issues within their supply chains, they still perceive the cost of deployment as too high:
As supply chains become more tangled, with a greater number of far-flung suppliers, managers are faced with risks that can crop up in dozens of countries. Companies have long used complex data sets to plan manufacturing to meet customer demand. But firms are now looking to combine data from external sources to better predict future risks.
Data science and domain expertise are not mutually exclusive: Some analytical skills are more important than others for data scientists who focus on SCM. And when it comes to training researchers and practitioners, expanding breadth of knowledge to include data science and functional business understanding is more important than depth in quantitative skill alone.
The researchers suggest the ability to apply predictive analytics techniques are, in a sense, more relevant than theory. For example, in statistics, a broadawareness of many different methods of estimation and sampling is more important than derivations of methods and proofs of maximum likelihood estimation. In forecasting, it’s more important to understand the application of qualitative and quantitative methods, rather than the underlying so-called stochastic processes. And in applied probability, it’s more important to use theory with actual data than to understand the theory of those stochastic processes.
That doesn’t mean theory doesn’t apply: Strong theoretical knowledge is crucial in SCM, the researchers say, along with the ability to apply analysis techniques from a broad variety of quantitative disciplines, particularly now that data and data variety are proliferating. What’s so important about theory? Using the appropriate logic and/or theory to build models prior to running predictive analytics is key to mitigating a proliferation of false positives, which result in wasted time and money.
But what are the practical applications of big data and predictive analytics that can be used by today’s managers? The researchers list a variety of processes that can be utilized in the three major links in the supply chain: manufacturers, carriers and retailers across forecasting, inventory management, transportation management and human resources.