It is no coincidence that Matthew Slifko’s research in predictive modeling in the presence of big and/or messy data deals specifically with the real estate market.
“As I prepared to return to grad school in July 2013, I was selling the house that I had bought at the beginning of the housing bubble in 2008,” said Slifko. “While predictive modeling had always been a favorite topic of mine as a student, real estate was a personal interest before it became an academic one.”
As a DAC Ph.D. student majoring in statistics, he was able to combine the two into a perfect fit.
Slifko said the type of data he works with can be problematic for a number of reasons. “For example, the size of the data presents computational challenges. And, incorrect data — such as a 100 square foot house with three bedrooms — interferes with the ability to build predictive models,” said Slifko, whose advisor is Scotland Leman.
“My research focuses on methods for using information about properties, real estate transactions, and market events like a natural disaster or a housing bubble to understand the behavior of property values in the presence of messy data,” he said.
Being a DAC student has given him the opportunity to collaborate with people from disciplines outside his own. “Learning how other disciplines view problems and how to communicate with non-statisticians is invaluable,” said Slifko, who is also a National Science Foundation research trainee in the Urban Computing (UrbComp) Certificate program, administered through DAC.
Earlier this year he was part of an UrbComp team that took second place during the final round of the 2018 Data Ethics Case Competition sponsored by the Center for Business Intelligence & Analytics.
Slifko earned a bachelor’s degree from the University of Pittsburgh at Johnstown and a master’s degree from Indiana University of Pennsylvania, both in math. His projected graduation date from Virginia Tech is Summer 2019, after which he hopes to secure an academic position.