Didja Know… Zillow is Better than Eyes!
Appraisers, I was busy churning out reports, and didn’t have time to send out this bit of news about Zillow earlier: Zillow is watching: New algorithm uses photos of your home to check quality and curb appeal.
Zillow is now claiming “it” can discern QUALITY of a home, just by looking at various photos they obtain from various sources! They claim that since QUALITY can be determined by its algorithm of looking at one dimensional photos, it’s Zestimate will be now much more accurate in determining three dimensional details… in fact, down to 2% accuracy!
Excerpt from the article:
“We’ve taught the Zestimate to discern quality by training convolutional neural networks with millions of photos of homes on Zillow, and asking them to learn the visual cues that signal a home feature’s quality,” Stan Humphries, Zillow’s chief analytics officer & chief economist, said in a Medium post announcing the new algorithm. “For instance, if a kitchen has granite countertops, the Zestimate now knows — based on the granite countertop’s pixels in the home photo — that the home is likely going to sell for a little more.”
Do any of you really wonder about this boast?
I do know one thing: training computer algorithms to understand property features is very difficult and time consuming… by humans making decisions… especially to get it right. I have seen evidence of how that works. It’s not easy to do.
But apparently, Zillow’s team is able to quantify the photo differences between laminate counters designed to look like granite, and actual granite… and apparently other items.
They claim accuracy is based on photos. What about the millions of homes (like mine) which have never had interior photos posted anywhere? I still live in the home we built in 1981. How in the heck is Zillow supposed to discern QUALITY of these homes, and then magically produce a presumed value, when there is no actual data upon which to base a decision?
Maybe it’s done with smoke and mirrors, rather than eyes?