An Analysis of Ride Hailing Preference Among University Students Using Artificial Neural Networks (ANN)

Abstract

The use of ride hailing has been growing rapidly, particularly in the post-Covid era. University students travel preferences differ from general public as they often adjust their travel according to lectures times. Therefore, it is important to investigate what drives students to use ride hailing service. This study fed on a stated preference survey among university students in Banda Aceh, Indonesia. Artificial Neural Networks (ANN) were utilized to establish links explaining relationship between travel mode preference and students’ socio-economic and travel characteristics. The lack of interpretability, as most machine learning techniques are notable of, were compensated with the use of variable importance. ANN models found that senior year students (5th semester or later), female students, and owning mulitple motorcycles increase the likelihood of using ride hailing service, while trip to campus, number of family members greater than three, and travel time less than or equal to 10 minutes would give minimum impact on the ride hailing use likelihood. ANN models exceeded the conventional logistic regression model in accucarcy testing by almost 28%. The findings might be used as guides in adjusting policy and operational system of the ride hailing services.