evaluating the effectiveness of burr mixtures for travel time reliability analysis

Travel time reliability plays a key role in travelers' decision-making processes and serves as a significant indicator for evaluating road performance. the initial stage of enhancing travel time reliability involves measuring it, which guarantees on-time arrivals and minimizes travel costs. a thorough understanding of travel time distribution is required to accurately measure travel time reliability. traditionally, travel time has been represented using unimodal distributions. however, recent studies have indicated that travel time distribution is often multimodal. failing to accurately model travel time distribution can lead to overestimation or underestimation of travel time reliability, resulting in suboptimal solutions in practical applications. this study mainly focuses on examining the effectiveness of six different mixture distributions, namely burr mixture, gamma mixture, inverse gaussian mixture, log-normal mixture, normal mixture, and weibull mixture, in modeling travel time reliability. the bayesian information criterion is utilized to compare and select the best model for representing travel time distribution. the results of this study indicate that the burr mixture model can characterize the travel time distribution more accurately compared to its alternatives. furthermore, the results show that using burr mixture distributions leads to more accurate estimations of travel time reliability measures compared to single distributions.