PREDICTION OF SCHOOL TRAVEL MODE CHOICE AND ITS DETERMINANTS – A MACHINE LEARNING APPROACH

Prediction of travel mode choice (TMC) is crucial for urban planners and policymakers to promote sustainable transportation systems and reduce traffic congestion. In recent decades, the prediction of TMC to schools, which involves daily commuting, has attracted the interest of researchers in green urban planning and a better society. Statistical models are based on many unrealistic premises about the data distribution and are typically used to perform mode choice analysis, which might result in biased model predictions. Moreover, machine learning algorithms that are assumption-free can handle complex, imbalanced, and multiclass datasets with high interpretability and outperform conventional techniques; thus, they have received much attention. Therefore, the present study intends to use modern techniques, such as Naïve Bayes, random forest, gradient boost, support vector machine, and linear regression, to predict the TMC to school (highest level of education) and its influencing factors. The current study contributes to the existing literature through (1) the application of modern techniques for the prediction of school TMC, (2) feature importance to predict the most significant feature of school TMC, (3) a proposal of the best predictive model, and (4) a discussion of the effectiveness of modern techniques over traditional methods. A total of 2756 samples from the NextGen 2022 National Household Travel Survey – California dataset was utilized to predict school TMC and its influencing factors. Based on the predictions, it was found that gradient boost outperformed other machine learning models with an accuracy of 98.9% in training and 83% in testing. Moreover, random forest achieved an accuracy of 77.8% and 71.1% in training and testing. Based on the sensitivity analysis, it was found that age is the most significant factor in determining the TMC to school, followed by the type of school. The findings will help policymakers and can be used to better understand modeling TMCs to schools, promoting sustainable transportation options.