The method, published in the journal Expert Systems With Applications, uses functional graphs to interpret the operation of the bus service. Unlike traditional models that prioritize physical proximity between points, this technique establishes links based on the similarity of speed patterns recorded in different areas.
To develop this tool, more than 50 million GPS records were processed. The data were organized using the H3 spatial indexing system, allowing for the comparison of geographical representations with functional relationships derived from the dynamic behavior of traffic on the island.
The results indicate that more than 85% of the connections detected by the functional graph link areas located more than five kilometers apart. This suggests that the system's operation is conditioned by shared dynamics between distant points, going beyond the logic of geographical proximity or belonging to the same route.
The research has also demonstrated its effectiveness in the automatic detection of incidents. The system identified an alteration episode in the TITSA network that occurred between November 6 and 10, 2024, showing a high correlation with traditional geographical data.
