Where does NYC's 311 system fall behind?
A relational database and Power BI dashboard built on real NYC 311 service request data, with an automated Python and SQL pipeline behind it.
The question
NYC's 311 line handles millions of service requests a year, everything from noise complaints to broken streetlights. Response times vary a lot by agency, borough, and complaint type, but that variation is buried in raw city data. This project turns it into something a decision maker could actually act on.
This analysis asks which NYC agencies actually meet a 72-hour response window on 311 requests, and how wide the gap is between the agencies that do and the ones that don't.
How it works
Pull
A Python script calls the NYC Open Data (Socrata) API for 311 requests.
Automate
A scheduled GitHub Actions workflow runs that script weekly, no manual step required.
Model
Data lands in a PostgreSQL star schema: one fact table, four dimension tables.
Analyze
SQL with CTEs and window functions answers questions like agency ranking and month over month change.
Visualize
Power BI connects to the database directly and models relationships, DAX measures, and the dashboard below.
Data model
Key finding
NYPD closes 99.9% of its 311 requests within 72 hours. Housing Preservation and Development closes just 25.3%, a 74.6 percentage point gap between the fastest and slowest agencies in the dataset.
Live dashboard
