Visual Analytics for Spatiotemporal Anomaly Explanation

Interactive system to detect and explain anomalies with STARMA, d3.js, and contextual event signals.

This project developed a visual analytics system to explore and explain spatiotemporal anomalies in crime data.

What Was Built

Spatiotemporal graph modeling of crime records.
STARMA-based anomaly detection workflow.
Interactive front-end in d3.js for anomaly exploration.
LLM-assisted external context retrieval to connect anomalies with events and news.

Outcome

Research article accepted at SIBGRAPI 2025.
System used for visual inspection and explanation of anomalous behavior over time and space.

Timeline

February 2025 - August 2025.