Boston Building Heating Systems: BCAN Data Integration

Created a building-level dataset for heating-system inference by integrating three City of Boston public datasets using Python and pandas, with parcel ID matching and address standardization to reconcile records across sources that don't share a common key.

Screened unstructured permit comments with a hybrid of keyword rules and API-based LLM prompts, surfacing records tied to heating installation, replacement, upgrade, and fuel conversion — turning free-text permit notes into structured signals.

Designed a client-facing Boston property map with search, filters, and exportable results, helping the Boston Climate Action Network (BCAN) identify likely high-emission heating systems and target buildings for decarbonization outreach.