A Random Effects Model for Travel-Time Variability Analysis Using Wi-Fi and Bluetooth Data

Abstract

This study examined variations in travel time through a random-effects model using Wi-Fi and Bluetooth data. Previous studies mainly divided a day into discrete time windows and estimated an individual distribution function for each window. However, this overlooks the interrelated nature of travel time among the different time windows of a day. The present research suggests a conditional random-effects model that portrays time-varying travel time as a continuous phenomenon. It was shown that the lognormal probability distribution is the best-fitted unimodal distribution. It was hypothesized that the conditional model with a lognormal distribution as the best-fitting choice associated with random effects, it better fits real data than do existing models in the literature. The proposed model demonstrated a strong relationship between the mean and the standard deviation of travel time, and between these two factors and time of day. The travel time data obtained from Wi-Fi and Bluetooth detectors on a highway section of Tehran, Iran were used as a case study.