Understanding tourists’ behavior toward transport mode choice by using machine learning methods

Abstract

Recently, transportation problems in evolving cities have become increasingly complex thus requiring innovative solutions. The complexity arises from the combination of urbanization, technological advancements, and diverse demographics. Therefore, integrating advanced methodologies, such as machine learning (ML), into transportation planning becomes essential. This study explores the application of ML techniques to predict and understand tourists’ transport mode choices when engaging in leisure activities. A comparative analysis of various ML algorithms including Neural Networks, k-Nearest Neighbor (k-NN), Naive Bayesian, Decision Tree, Support Vector Machines, Discriminant Analysis, and Ensembles reveals distinctive strengths and trade-offs. Based on the results, Neural Networks demonstrate high accuracy, precision, recall, and specificity, which makes the algorithm well-suited to predict transport mode choices. While k-NN exhibits competitive precision and recall, it struggles with specificity and indicates a higher false positive rate. These findings support transportation planners to select the most suitable algorithm according to the desired outcomes. The adoption of these advanced approaches enhances the understanding of travel behavior thus offering valuable tools for developing efficient urban transportation systems.