Understanding the Travel Behaviour by Gender: Comparisons of Machine Learning Models

This study is conducted in Kuantan City to investigate the travel behaviours of Kuantan travellers while choosing their daily travel mode. In this study, five (5) machine learning models, namely, Neural Network Logistic Regression, kNN, Random Forest, and SVM are adopted to classify travel behaviour according to gender. The total of 386 respondents are given six (6) trip scenarios that consist of the 13 variables using RPSP Survey. The questionnaires can be divided into two categories (a) socio-economic characteristics and (b) travel behaviour characteristics. Feature importance technique is employed to rank the selected features according to the most important until the least important features. For male travellers, the most crucial features indicated by waiting time, region, and walking distance from the last stop to the destination. The least important features depicted by male travellers are employment status, DOM, and income. Meanwhile, the features for female travellers show that the most crucial features indicated by region, waiting time and total travel time. The least important features depicted by female respondents are age, walking distance from home to nearest bus stop, and employment status. Among these models, Neural Network depicted the most accurate model to classify the respondents’ travel behaviour.