A competition aimed at improving the Zillow Zestimate home-valuation algorithm attracted more than 3,800 teams when it was announced last May, and now the Seattle-based real estate tech company is revealing the top three teams to emerge from round one.
As part of the $1 million Zillow Prize, the three teams will be taking home $50,000 collectively and be part of the group of 100 teams that move on to the second round and compete for the grand prize.
Launched in May 2017, Zillow said in a news release Wednesday that the Zillow Prize has “become one of the most popular machine learning competitions ever on Kaggle,” the platform designed to connect data scientists with such problems.
Here’s more on the top teams so far:
First place, $25,000: Team Zensemble
The team is made up of participants from three different countries — Russ Wolfinger (United States), Dmytro Poplavskiy (Australia), and Jonathan Gradstein (Israel) — who first met online as competitors in another contest.
Wolfinger told Zillow that he plans to donate all his winnings from the first round to the math and science program at Meredith College, a women’s liberal arts school in North Carolina, where his wife is the dean of Natural and Mathematical Sciences.
Poplavisky said that his hard work on Zillow Prize caused his electric bill to double because of the uptick in server usage. He installed solar panels on his Brisbane, Australia, home to offset the cost.
Second place, $15,000: Team Silogram-2
The father-daughter team of Phil Margolis, an independent consultant focusing on machine learning contract jobs, and Isabel Margolis, a 21-year-old student studying math and computer science, worked on the competition from their home in Zurich, Switzerland.
“I’ve bought and sold a home in the U.S., and you look at properties and there’s all these theories about how to valuate them,” Phil Margolis said. “This was really an interesting case where we could apply really advanced machine learning technology to real world data.”
Third place, $10,000: Ryuji “Jack” Sakata
One of many solo competitors chasing the prize, Ryuji Sakata is resident of Osaka, Japan, is a 30-year-old data scientist who has spent the past six years at Panasonic working to improve manufacturing using data analysis. He has never visited the U.S.
“I am not familiar with the U.S. housing market, but a challenge of creating a real forecast for the future was very interesting,” he said.
Zillow has more on the finalists competing for Zillow Prize here.
“We’ve been blown away by the data science community’s response to the Zillow Prize. There were lots of innovative solutions — and hundreds of teams were able to make incremental improvements in the Zestimate’s accuracy,” said Stan Humphries, creator of the Zestimate and Zillow Group chief analytics officer. “The Zestimate is trying to answer an incredibly complex and important question — how much is my home worth — and we’re so excited to see what new innovations these top 100 teams bring to the Second Round.”
The second round of the competition opened on Feb. 1. The winning team must ultimately build an algorithm to predict the actual home sale price itself, and home value predictions from each algorithm submission will be evaluated against real-time home sales in August through October of this year.
“To win the $1 million grand prize, an algorithm must beat Zillow’s benchmark accuracy on the Second Round competitions data set and enhance the accuracy further than any other competitor,” Zillow said.
The Zestimate’s current margin of error is 4.3 percent, nationwide.
A $100,000 second place prize and $50,000 third place prize will also be awarded.