Geek Alert—Zillow Announces $1 Million Competition To Best Its Tech Team
Zillow wants to make its Zestimates more accurate and is holding an open call for data scientists everywhere to get involved.
Calling all geeks and tech nerds. Zillow has decided it needs to rework its highly contentious, Zesitmate algorithm and it’s willing to pay the winning person or team $1 million to do so. In an open call that is a cross between Charlie & The Chocolate Factory and Ex Machina, the winning algorithm will be tested in real time to see if it can help improve estimating the value of properties across the nation.
For the past decade Zillow has hung its hat on its ability to offer free comps via its Zestimate feature but buyers and sellers alike have been frequently disappointed by its wildly inaccurate numbers. Issues algorithm based estimating face are not unlike algorithm based selections with music streaming. They miss the nuances. In real estate, not being able to take into account block to block fluctuations in a neighborhood or if a property is in need of repair can greatly change the value.
Although Zillow has always maintained that their Zestimates are just one data point of reference to be used in conjunction with other information such as a realtor’s expertise, many people who use it do so expecting greater accuracy. If the company’s quest to enhance the algorithm works it would go a long way to making the site indispensable. As it stands its more noted for the conjecture it creates rather than pin point numbers. Zillow co-founder and executive chairman Rich Barton, also held a similar competition at Netflix, (he’s on the board) board he sits on but never implemented the winner’s algorithm. He called the Zestimate “very provocative and personal and a little voyeuristic” in a 2016 GeekWire interview discussing how the company came up with the tool.
Zillow stated that the The U.S. median absolute percent error (measured by the eventual sale price of a home) currently stands at five percent, improved from 14 percent in 2006.
“We still spend enormous resources on improving the Zestimate, and are proud that with advancements in machine learning and cloud computing, we’ve brought the error rate down to 5 percent nationwide,” said Stan Humphries, Zillow Group’s chief analytics officer and creator of the Zestimate on Zillow’s Media Room.. “While that error rate is incredibly low, we know the next round of innovation will come from imaginative solutions involving everything from deep learning to hyperlocal data sets — the type of work perfect for crowdsourcing within a competitive environment.”
Zillow announced that their “Zillow Prize” hopes to attract data scientists everywhere to try their hand at tinkering with “one of the highest-profile, most accurate and sophisticated examples of machine learning.”
For serious tech & math wiz’s spurred by the thought of $1 million in prize money Zillow offers more information on their challenge in their media room. The ultimate challenge will be a private affair where the winner from the qualifying rounds will go head to head with Zillow’s own team.
“To take home the $1 million dollar grand prize, the winning algorithm must beat Zillow’s benchmark accuracy on the final round competition data set and enhance the accuracy further than any other competitor,” Zillow’s site says. “A $100,000 second place prize and $50,000 third place prize will also be awarded in the final round. A total of $50,000 will also be awarded to the top three ranking teams in the qualifying round.”
Let the battle commence!
AGORAFYAbout 65 percent of the units are from buyers, most of whom are native New Yorkers – plus many already live in the neighborhood. #NYC #21East12th https://goo.gl/ZupI5x
AGORAFYIncreasing inventory is an important contributor to improving the affordable housing crisis in the United States. #AffordableHousing #Apartments https://goo.gl/sYOHbj
AGORAFYWhile the rich park their cash in empty condos, creatives are being squeezed into different neighborhoods and cities. #Condos #Creatives https://goo.gl/mDrLU1