Over the last 15 to 20 years, companies have bought into the notion that they could use computing power, data and analytics to discern more about their customers, operational efficiencies and the markets. But, as they say, numbers don’t always tell the story.
Jeffrey T. Prince, professor and chairperson of business economics and public policy at Indiana University’s Kelley School of Business, says reasoning and logic are often left out of the equation.
“And that really should be the foundation for all of this,” said Prince, author of the new book, “Predictive Analytics for Business Strategy: Reason From Data to Actionable Knowledge,” published by McGraw-Hill.
He noted that predictive analytics frequently is used to track consumer patterns and then evaluate data to determine future consumer behavior. “That’s the creepy part of analytics, where we can see what you’re up to and make accurate predictions about what’s going on or what you’ll do,” he said.
It’s what Prince calls “passive prediction.”
“Active” versus “passive” prediction
In his classes at Kelley and in his book, he also presents the benefits of “active prediction,” focusing on predictions about how business strategies impact future business outcomes. He shows how these predictions can be used to make effective business decisions. He writes about the need for good old perspective, reasoning, and understanding of the scientific method, and explains analytical methods within that context.
He writes, “predictive models do not always yield perfect predictions. These models rely on assumptions, some testable and others not. However, they provide a structured mechanism that allows us to use what actually has occurred (recorded as data) to inform us about what alternative strategies will accomplish.”
It’s not entirely new. For decades, econometrics has been used to analyze data to help leaders anticipate future trends. But as a group, business economists have become reticent to use the words “predict” and “forecast” to explain their projections.
Prince, who also holds the Harold A. Poling Chair in Strategic Management and co-directs Kelley’s Institute for Business Analytics, “absolutely” embraces these terms.
“I think it’s been a huge mistake by economists to draw these silly lines about semantics, while we watch other fields blow by us in this area and get relegated to irrelevance,” he said. “I looked it up on Google Trends. If you go from 2004 to the present and you type in ‘econometrics,’ it craters. If you look up ‘predictive analytics,’ it’s through the roof.
“That’s not a coincidence, and I think that a big problem that we’ve had is that we’ve allowed ourselves to get cut out of the conversation when it comes to predictive analytics,” Prince added. “The point I’m trying to make in the book is what we do is prediction with a very general definition of what prediction is.”
Webster’s Dictionary defines prediction as “a statement about what will happen or might happen in the future.”
Some business disciplines have defined passive prediction as the only type of prediction being used today within companies. Prince said his book received pushback from a few peers who take this view.
“If you look at the definition of prediction, it’s so straightforward. I don’t know what could possibly be controversial about it,” he said. “If you buy into the definition, then what we’re doing as economists, when you think about econometrics trying to explain the effects of policy changes and strategies, those are absolutely predictions – active predictions.”
Prince recently gave a talk about this topic and his book at the American Economic Association, where a CFO told him afterward, “we are in desperate need of this because people don’t know this stuff…. Why? Because they’re all doing data mining.”
If they are going to advise on decision-making or actually make the decisions, graduates at Kelley and other business schools need to understand what kind of information to rely on, said Prince, because having an understanding of only passive predictive models will lead to a loss for their companies.
“The reality is that both approaches have to use historical data,” he said. “With passive prediction, you’re saying this is what’s going to happen because this is how all of these variables moved together. With active prediction, we’re interested in how variables move together, but we’re also zeroing in on particular types of movement that have implications about effects.
“This allows us to not only think about how the variables might move together in the future, but how a movement of one variable will cause a movement of another variable, and that’s a key distinction.”
Prince’s earlier book with fellow Kelley business economics professor Michael Baye, “Managerial Economics and Business Strategy,” was designed to teach managers the practical use of basic economic tools and applied them to real-world examples. Prince has been teaching at Kelley since 2010 and has been using material from the book in classes for the last five years.