SLA compliance and response variability analysis using NYC 311 service requests
A decision-oriented BI project analyzing SLA compliance and operational performance using real NYC 311 service request data.
Built with API-driven ingestion and a modern analytics-first architecture.
To evaluate operational efficiency and SLA adherence across NYC agencies using real, API-sourced data.
The project is intentionally framed around analytical rigor, pushing heavy logic upstream and keeping BI tools focused on decision-making visuals.
This project prioritizes analytical correctness and governance over visual storytelling.
- Platform: NYC Open Data (Socrata)
- Dataset: 311 Service Requests
- Dataset ID:
erm2-nwe9 - Access Method: Socrata API via Python (
sodapy) - Data Scope: Time-scoped operational slice (explicitly framed, not full historical analysis)
- Raw data is ingested via authenticated API access and stored without transformation.
- Raw data stored in Parquet for columnar, compressed analytics
- Raw files treated as immutable
- CSV-based workflows intentionally avoided
- This ensures raw data can always be re-audited against derived analytics.
- DuckDB used as the analytical SQL engine (OLAP)
- Parquet queried directly without import
Business logic is implemented upstream using layered SQL views:
- Raw views: faithful exposure of ingested data with no interpretation
- Normalization views: controlled semantic mappings (e.g., status normalization)
- Eligibility views: enforcement of SLA applicability rules
- SLA views: fixed-threshold SLA classification and severity buckets
Power BI consumes only curated views and does not define eligibility or SLA logic.
This project is designed to be fully reproducible from raw data ingestion to BI-ready analytics.
The intended workflow will include:
- Environment-based API authentication
- Deterministic data extraction to immutable Parquet
- Automated construction of the DuckDB semantic layer via SQL scripts
- Read-only BI consumption via ODBC
Detailed, step-by-step reproduction instructions will be added as the pipeline stabilizes.
Rehan Abdul Gani Shaikh
Data Science & ML Student | Python • Power BI | Building Real-World Data Projects
🔗 Connect with me: LinkedIn
📬 Email: rehansk.3107@gmail.com