Resumen
Speed variations are widely believed to be one of the major contributing factors to the occurrence of accidents, which is a critical problem related to the traffic safety. This problems becomes even more challenging for highly congested roadways such as I-95, Florida's Turnpike and US-1 highways located in the Southeast Florida, which are not only known as very busy roadways but also listed as having high fatality rates due to accidents. This paper attempts to study and analyze the speed variations on these inter-urban parallel highways with a focus on the Thanksgiving holiday, one of the most travelled holiday periods of the year in the U.S. As a part of this evaluation, northbound and southbound speed patterns are analyzed for the 2012 and 2013 Thanksgiving holidays based on the following clustering approaches: Gaussian Mixture (GM) with classical and greedy EM, K-means and X-means algorithms. Special consideration is given to accurately identify the low and high speed regimes on the roadways, which is critical to provide a better understanding on the crash occurrences and the congestion characteristics. Transportation agencies can use this information to implement appropriate traffic-related countermeasures before and after the holidays that can lead to a reduction in accidents and delays.