Case Study: BuildFax achieves fast, property-specific roof-age and cost predictions with Amazon Web Services (Amazon Machine Learning)

BuildFax - Customer Case Study

BuildFax, founded in 2008 and based in Asheville, NC, aggregates building-permit and property data to help insurers, inspectors and analysts make better decisions. Facing billions in annual roof-loss payouts, the company needed faster, more accurate, property-level roof-age and cost estimates—its prior ZIP-code–based models built with Python and R were complex, slow and offered limited differentiation.

By adopting Amazon Machine Learning, BuildFax trained models on tens of millions of known roof ages and property characteristics and delivers real-time, property-specific predictions via APIs. Model development time dropped from six months to under four weeks, customers receive more precise roof-age and job-cost estimates (text analysis yields about 80% accuracy), and BuildFax can now offer new programmatic analytics services.

BuildFax

Joe Masters Emison

VP of Research and Development