Alumnus Focus: Zillow Chief Economist “Zestimates” the Value of His Liberal Arts Education
Real estate is not rocket science.
Or is it?
Ask Stan Humphries '90. He's responsible for Zillow's "Zestimate," an estimated market value on every U.S. home, which is integral now to all things real estate-related on the internet since its introduction a few short years ago.
In the late '80s, Humphries was an aerospace engineering student at Georgia Tech. He loved the academic work, but as time passed he found he did not want to become an engineer after all. He transferred to Davidson, studying political science and economics through an interdisciplinary major, with an eye toward science and technology policy.
The Davidson years were formative, personally and professionally. He met his future wife, Katherine Bagby Humphries '90.
"It's not just a cliché, what they say about the liberal arts," said Humphries. "In my case, it gave me a way of thinking about the world and a critical faculty for thinking about issues and breaking down problems. It also gave me an enlarged worldview in terms of what I should be thinking about. I left Davidson thinking about life being a continual learning exercise."
After graduation, he taught chemistry and physics in the Peace Corps in Benin, West Africa. There, he learned about cultural assimilation and identity and privilege and conflict, as Benin's peaceful but nonetheless disruptive transition to democracy at the time often interrupted his teaching.
"I think the whole experience lowered my perceived barriers for a lot of opportunities that I might want to take on," he recalls of that time spent alongside mostly French and German volunteers.
Next he went to Washington, D.C., where he earned a master's degree in foreign service at Georgetown University while working as a research assistant at the Brookings Institution. Then, as a Presidential Management Fellow with NASA as his home office, Humphries witnessed the development of international and interagency agreements for different space activities. And he got to rotate around as a Presidential Fellow, working for a time in the Technology Administration at the U.S. Department of Commerce and then for the president's science advisor in the Clinton White House.
"It was a huge amount of fun," he said.
Next came the University of Virginia for a Ph.D. doctorate in government (subfields: quantitative methodology/American politics), as well as professional stints in the university's classrooms and research computing lab. During the course of his graduate work, he became more interested and involved in the methodology and techniques of data analysis, not just the substantive issues of policy and economics.
On the Forefront
By the time he ended up in Seattle in 2001, Humphries was primed to deal with big data.
"With the advent of the Internet, there were opportunities to apply the econometrics skillset to Internet data," he said. His first West Coast job was at Expedia. "Those problems were initially around customer segmentation and relative value of customers. It was in the early stages of [Internet] personalization."
Soon he was managing the customer analytics team at Expedia, applying predictive models to figure out which products to show to customers and at what price. The team also forecasted marketplace supply and demand at, for example, Las Vegas's famed Bellagio, Venetian and Caesar's Palace casino resorts.
Then came a ground-floor opportunity as part of the nascent Zillow in 2005, where Humphries was soon to lead the rapid-fire development and deployment of the now ubiquitous Zestimate.
At first, many of the challenges at Zillow were more basic and logistical, like collecting all the data from 3,000-plus counties across the United States. It was labor-intensive, but not necessarily complex.
"The benefit we had in 2005 is that most people predicting home values had used relatively simple techniques across large geographies, maybe a simple linear regression model for a whole state," Humphries said.
Then, things got more complex – both challenges and solutions.
"Early on, I brought in algorithms from completely different domains, for instance genomics, biotechnology, and image compression for digital photography," he said. "Researchers in these fields were solving problems that looked like problems you also have in real estate. Then came more machine learning techniques and algorithms at micro-levels of geography, down to individual ZIP codes."
In 2006, the Zestimate launched.
From Zillow's director of advanced analytics, Humphries progressed to vice-president for data and analytics, then to his current post as chief economist.
"Now I spend a lot of my time doing economic research and writing, and it brings it back to a lot more of what I did at Davidson. Communications is a huge part of my job now," he said. "A Davidson education has allowed me to traverse the line between technology and non-tech business areas."
So: rocket science or real estate?
"It's all about data, information and insight," said Humphries. "Sales prices are data. Sales prices for last week are information. And insight is what that means for the worth of an individual home, now or a year from now.
"We really focus on the sophistication of how it's all applied. It's not just the size of the data set, it's what you do with it."