Estimation of travel time of dolmus as a paratransit vehicle with artificial neural networks: The case of Ortahisar (Trabzon)

The aim of the study is to estimate the travel time of the dolmuş, which has an important role in public transport, has a complex, irregular and difficult to understand structure, by taking into account land use decisions, using artificial neural networks method. It was decided to use artificial neural networks to identify the functional relationships between inputs in the travel time estimation of dolmuş systems, which have a complex structure. Three different dolmuş stops on the east, west and south directions of the city were selected as a sample, taking into account the operating order, trip frequency and the routes. In the study, different numbers of input variables were tested in artificial neural network models to estimate the travel time. The success of the estimation model was analysed when the inputs related to land use were used alone or in combination with other inputs. Although the estimation was not successful when the inputs related to land use were used alone, the model performed the best estimation when the input variables "congestion" and "average speed" were used together with land use.