Analytics yield loss – from research to use

0
27 views

Share on LinkedIn

Why is there substantial degradation of academic research in converting its findings in applications such as business, health care, and government? I ask this question because this week I am attending the annual conference of the Institute for Operations Research and the Management Sciences (INFORMS) being held in Charlotte, North Carolina. There about 4,000 in attendance. Similarly each year I attend the annual conference of university accounting faculty and doctoral students at the American Association of Accountants Management Accounting Section.

Every time I attend events like these I am awestruck by how rigorous researchers have investigated a problem. However, I am also disillusioned by what appears to me to be a low level of application of the insights and ideas that could and should come from the research.

Too theoretical or too close minded?

There is decay and degradation in the physical world. Most food products are perishable with time. There is loss in electricity power from its generation source during distribution across the grid. Many manufacturing processes experience yield loss with scrap and waste. However, science has addressed these to minimize the yield loss – preservatives and refrigeration for food, wiring materials for electricity, and quality improvement techniques in manufacturing. Why can’t there be ways to get more yields from academic research?

One barrier might be that some of the research might appear to be too esoteric or incomprehensible to practitioners. With operations research, statistics, and analytics the terminology can be somewhat intimidating. Examples include “group heuristics in multi-attribute choice environments”, “dualization to solve stochastic optimal control problems”, “integer linear programming reformulations”, and “multivariate adaptive regression splines.” But before you dismiss this type of language as overly theoretical, consider that the research addresses relevant and important needs including electric vehicle battery life, supply chain management, health care delivery, freight transportation, route optimization, and genes and disease management.

My belief is that part of the problem is on the receiving end. Organizations do not sufficiently invest the time to understand and evaluate the massive amount of research that is conducted and available.

It should be a two way dialog

I am not letting the researchers off the hook. Could they more effectively communicate and distribute their research to organizations that could potentially benefit? Of course. But on the receiving end organizations could be more open-minded and consider a longer term view at how they could apply the research.    

An inspirational and motivational plenary session talk at the INFORMS conference was by my co-worker, Keith Collins, Senior Vice President and Chief Technology Officer at SAS. His presentation, “The Evolution of Analytics”, explained how the exponential growth in data will require a shift in IT thinking from putting “data in” storage to how to get decision-relevant “data out” and easily accessible. As data exceeds storage capacity, IT will need to get cleverer to exploit it. Keith described three next future advances: (1) how advances in high performance computer technology are combining with progress in analytical thinking; (2) how visualization is not sufficient without analytics to make sense of it; and applying text analytics to unstructured data.

With time and better collaboration of business and government with researchers and academia, the yield loss from analytics can be minimized and knowledge transfer maximized.

LEAVE A REPLY

Please enter your comment!
Please enter your name here