Analaysi̇s of travel time variability and prediction of travel time using artificial neural networks

Analysis of travel times helps people determine their travel behavior. Knowing the variability and reliability of previous travel times when going from one place to another can make good use of time. In this study, the variability and reliability of the travel times of the journeys made by taxi between the La Guardia airport of New York City and the John F. Kennedy airports were investigated. Two methods were also used in the study to estimate the travel time. One of them is the nonlinear autoregressive model -NAR method, which uses only the travel time, and the other is the nonlinear autoregressive model with external input -NARX, which also uses temperature, visibility and wind variables, which are the factors affecting the travel time. Different training options were applied in the two methods and the best model for both methods was found in the 'sgdm' training option.