• BYU Professors Partner with Center on Pricing Research
• Advisory Board Member Honored by College of Physical and Mathematical Sciences
• Victory Seeker Finds Success
• Note from the Director

• Register for the FREE newsletter
• Submit ideas for future articles
• Become a partner with the eBusiness Center

• Newsletter Archives

• Contact the editor of the eBusiness Center Newsletter:

Rachel Finley
574 TNRB
Brigham Young University
Provo, UT 84602
(801) 422-3598
prspecialist@byu.edu

BYU Professors Partner with Center on Pricing Research

Patrick Mullen is used to tinkering with things: he began with his family’s first computer.

“It started as an interest,” he says. “I used to play computer games and the first computer we had barely had enough resources to run them. I would have to get into the computer and tinker with it.”

Mullen’s childhood programming penchant has led him to pursue a master’s in BYU’s computer science department, where he gets to tinker with more than just computers.

As part of his master’s thesis, Mullen and faculty member Kevin Seppi have altered a well-known algorithm that will better allow online businesses to map dynamic pricing and optimize their earnings.

The particle swarm optimization model compares market prices to particles and those particles are a bit like feeding animals, Seppi says. Like a flock of seagulls, particles, or prices, swarm to prime eating grounds and stay there until a more plentiful feeding ground becomes available.

All businesses are looking for that “good feeding ground” the place where they make the most profit, Seppi says. Finding it means testing-the-waters, setting experimental prices and monitoring market reactions. But experimentation often means losing time—and money.

“Businesses don’t want to lose money,” he says. The bottom line is usually easier to find; it’s the upper limit that a little vaguer. “Dynamic pricing is a problem for someone who sells a commodity or set of commodities. They need a mechanism that allows them to set a price that maximizes revenue.”

Mullen and Seppi have implemented a modified version of the particle swarm optimization model that incorporates the unpredictable into determining optimal prices.

“There are several predictable factors that influence demand for a product,” Seppi says. “For example, when it’s hot outside, more people want to buy ice cream.” Such constants can be easily included in a set algorithm. But other factors—factors that are not so visible—cause anomalous readings and skew the results.

That’s where Mullen and Seppi have added an extra dimension to the model to allow for environmental noise and market change. Essentially, they are teaching the algorithm to think and learn as the market moves around it.

“We’re happy to sponsor multi-disciplinary research that brings in colleagues outside the Marriott School,” says Stephen Liddle, director of the Rollins Center for eBusiness. “E-business is a topic with wide-ranging impacts on society, and we need various perspectives to address it well.”

Seppi focused his PhD work at the University of Texas—Austin on the idea of uncertainty. Working with Mullen on his master’s thesis has been an opportunity to continue some of that research. “They’re things you can’t really predict,” he says. “If you account for the uncertainty at the beginning you can sometimes get better results in the real world.”

“There are things we just don’t know about how the market is responding to our services,” Seppi says. “Maybe businesses know something about the market and there’s some intuition involved, but it’s great to have a second opinion.”

In their research, Mullen and Seppi tested the algorithm using five market patterns: a constant market, an up trend, a down trend, a random market, and an intervention, or suddenly changing market.

The algorithm performed well in four of the five categories—it was “quick on its feet,” Seppi says, meaning it adapted quickly to market fluctuation and learned to ignore pricing anomalies.

Allowing for noise in pricing research is a less-studied, but crucial, area of business.

“This is a fruitful place to look,” Seppi says. “Think about how many web sites there are that sell their products online. There are sites that sell so many products one person can’t monitor the pricing for all of them. Other sites sell one or two products, but it’s a side business, so the person isn’t involved in monitoring prices either.”

Seppi hopes that this research will not only help businesses, also consumers and the economy.

“We’re trying to find the most efficient way to deliver the goods and services to the market,” he says. “We try to set the price in the most efficient place that we can. That makes make our economy fundamentally more efficient, and that’s good for everybody. Although it may not be worth a human’s time to look for that efficiency, it may be worth a computer program’s time.”