This new startup is bringing predictive data science to real estate.
We live in a world where complex predictions can often be expressed with a single number. Banks use FICO scores as being predictive of whether you will repay a loan on time. Universities use SAT scores as (supposedly) being predictive of how you will do in college. Nightclub owners use a band’s social media numbers as indicative of whether they should book the band (seriously).
Peter Lufkin, a data scientist in Santa Barbara, has developed a system to predict the condition of real estate property without a physical inspection. If you’re looking at buying a house somewhere you can look up its predicted selling price using a site such a Zillow, but there’s no way to know its condition and potential deferred maintenance without sending a contractor out for a physical inspection. Peter has founded a new startup, Pomar Lane, to commercialize his product, which is called the Pomar Condition Scoring System.
In a pilot project in one large county in California, the Pomar parametric models were solidly predictive (r=.8) of home condition, at a dramatic cost savings over expensive physical inspections. The Pomar System can reduce insurance claims, improve investor decisions, and make home valuation models much more accurate.
The potential users of the Pomar System are broad — from portfolio owners to individual home buyers, from insurance companies to financial services institutions. Three years into development, Pomar Lane is ready to go from their initial pilot project to national scale. I predict great success for them.