Exploring spatio-temporal traffic performance variation through clustering of descriptive travel time statistics

Characterizing links in road networks is vital for understanding recurring traffic state patterns. For long-term planning, clustering can reveal links with similar characteristics and patterns that may indicate degraded performance in the future. In this paper, we apply cluster analysis to automate this process and identify similarities among links and days to find potential infrastructure deficiencies. Our study uses different clustering techniques on descriptive statistics to categorize link-types and day-types. Applying our method to high-resolution travel speed data, reveals consistent link characteristics across different clustering algorithms. The preliminary results show that the identified clusters maintain stability both in space and time, confirming their effectiveness in identifying consistent link characteristics and daily patterns. This offers insights into traffic state variations based on travel speed.