Total Tap Events
â
Historical + Real-time Stream
Active BRT & Feeder Routes
â
GTFS Conformed Dimension
Transit Stops & Shelters
â
WGS84 Coordinates Mapped
Data Quality Gate
80 / 80
100% dbt Tests Passing
Hourly Tap-In Ridership Pattern
Peak AnalysisCard Issuer / Bank Market Share
Payment TelemetryTop 5 Busiest Boarding Stations
Shelter Demand| Rank | Stop / Shelter Name | Latitude | Longitude | Total Tap-Ins |
|---|---|---|---|---|
| Loading live warehouse telemetry... | ||||
End-to-End Enterprise Transit Data Architecture
Medallion architecture, DAG dependencies, and dimensional models.
1. Ingestion Layer
- GTFS Puller: Weekly Python pipeline downloading official TransJakarta transit feeds.
- Kafka KRaft: Real-time replay producer streaming transaction payloads to
taps.raw. - Dead-Letter Queue: Pydantic schema validation isolating malformed messages to
taps.deadletter.
2. Storage & Modeling
- PostgreSQL 16: High performance relational storage (schemas:
raw,staging,warehouse). - dbt Core: 11 models transforming raw entities into a conformed star schema (
fact_taps,dim_*). - Quality Gate: 80 automated unit & schema tests blocking corrupted data from reporting views.
3. Orchestration & BI
- Apache Airflow: Automated DAGs for weekly GTFS ingestion and nightly warehouse builds.
- Apache Superset: Read-only visual analytics with
superset_rosecurity roles. - Guest Token Gateway: FastAPI service issuing ephemeral guest tokens for public dashboard embeds.