How does it work?

parkUCF’s architecture is intentionally simple but robust. The system is built on a fully automated, cloud-native, end-to-end ELT (extract, transform, load) data pipeline designed to continuously ingest, store, and analyze parking occupancy data with minimal operational overhead.

1. An AWS EventBridge rule triggers a Lambda function every two minutes, matching the update interval of the UCF parking API. Each invocation queries the API endpoint and inserts the resulting snapshot as a new row into a PostgreSQL database.
2. At the end of each day, the database table is truncated and its contents are exported as a single consolidated JSON file to an Amazon S3 bucket. This preserves a lightweight historical record while keeping the transactional database small and efficient.
3. On a monthly schedule, AWS SageMaker Pipelines triggers a notebook job that trains a Facebook Prophet time-series model on the accumulated data. The resulting model outputs updated forecasts and summary metrics, which are then surfaced throughout the application.