Large-Scale Data Management for Property Characteristics Development
A leading property insurer in the United States is using the Descartes Labs Platform to derive insights from property characteristics that drive risk. While commercial sources of property characteristics exist, they take a “one size fits all” approach. Our customer wants to catalog attributes that they know indicate risk, and experiment with new and untested attributes quickly.
To create a living property attributes catalog, our customer sourced a large collection of aerial imagery in order to extract property information using computer vision. By starting with the imagery itself, the insurer is unconstrained in how attributes are defined. Even so, petabytes of geospatial data presents its own challenges. Geospatial imagery requires specialized processing and access patterns and significant computational resources.
Fortunately, the Descartes Labs Platform is a best-in-class solution to make geospatial data painless and accelerate data science innovation. First, the platform can ingest, pre-process, and render analytics-ready even the largest data sets in real-time. Second, our Workbench is a rapid-development environment providing visualization tools designed to surface insights in the data and facilitate collaboration between data scientists and business users. Third, operationalizing and scaling models nationwide is dramatically simplified due to our scalable compute architecture and real-time data ingest. Only the Descartes Labs Platform is built to discover and operationalize custom property attributes on a nationwide scale.