It is a seller’s market, and Josh Grubman ’03 (Boca Raton, Fla.) is examining ways to help sellers get top dollar for their homes as part of a year-long research project.
A mathematics and economics major, Grubman is conducting an honors thesis under the guidance of Thomas Bruggink, associate professor of economics and business.
Bruggink’s research has been published in numerous academic journals and books, including a forthcoming article in International Business and Economics Research Journal, and has been cited in many newspapers, including The Wall Street Journal and USA Today. He is a referee for 11 periodicals, primarily economics journals.
Grubman began his research by identifying 100 single-family homes for sale in Randolph, N.J., and recorded information pertaining to each — age, number of bedrooms and bathrooms, and price — on a computer spreadsheet. He is now using computer programs that apply multiple regression analysis and artificial neural networks, a system that mimics the way the human mind works, to learn which method predicts the sale price of homes more accurately.
“I’m using several different permutations of the data set as well as a number of measures of success in prediction,” says Grubman, explaining that while researchers frequently use multiple regression analysis to predict price in a number of areas, artificial neural networks are usually reserved for more technical applications.
Bruggink says he’s impressed with Grubman’s knowledge of artificial neural networks and by the fact that he searched the Internet for appropriate software to conduct the research.
“He acquired the neural network software on his own,” the professor adds. “I’m not sure what we would have done without it.”
Bruggink says the research could eventually contribute to a new method of setting asking prices for homes that would replace the approach of looking at sale prices of similar homes in a neighborhood.
Grubman decided to conduct the thesis research after writing a paper on housing prices for an econometrics class last year.
“I was interested in seeing how powerful the artificial neural networks technique is compared to the simpler multiple regression analysis formula we use now,” explains Grubman, who is glad to be working with Bruggink because “he knows a lot about the process, is very good at maintaining expected levels of progress, and ensures that I’m heading in the right direction.”
Grubman was a member of the team that won last year’s Math Bowl, a competition in which four squads squared off in three matches using a “Jeopardy”-style format. He also played a key role on the team that took third in the Team Barge Mathematics Competition.
Grubman, who plans to attend law school next year, completed an internship in personal automobile pricing last summer at Traveler’s Property Casualty in Hartford, Conn. “It was definitely a learning experience,” he says.
He is a member of Lafayette Environmental Awareness and Protection and has competed in intramural billiards and soccer. He is a member and former president of Phi Kappa Psi fraternity.