Predicting the use frequency of ride-sourcing by off-campus university students through random forest and Bayesian network techniques

This study used a survey technique to investigate factors that motivate the adoption and the usage frequency of ride-sourcing among students in a Malaysia public university. Two of the most broadly used machine learning techniques, Random Forest technique and Bayesian network analysis were applied in this study. Random Forest was employed to establish the relationship between ride-sourcing usage frequency and students' socio-demographic related factors, built environment considerations, and attitudes towards ride-sourcing specific factors. Random Forest identified 10 most important factors influencing university students' use of ride-sourcing for different travel purposes, including study-related, shopping, and leisure travel. These important predictors were found to be indicators of the target variables (i.e., ride-sourcing usage frequency) in Bayesian network analysis. Bayesian network analysis identified the students' age (0.15), safety perception (0.32), and neighbourhood facilities in a walkable distance (0.21) as the most important predictors of the use of ride-sourcing among students to get to school, shopping, and leisure, respectively.