Machine Learning-Based Travel Mode Prediction: A Comparative Methodological Approach

Since the last few years, cities are continuously expanding because of urbanization. This urbanization has increased the need of effective management of resources such as public transportation, private vehicles, roadways, and several other things. Forecasting and study of travel mode selection provide insights into the factors influencing individuals' selection of transportation modes for their trips. This analysis plays a significant role in promoting sustainable transportation in smart cities. Hence, in this work, prediction, and evaluation of travel mode selection is carried out using different machine learning algorithms. The comparative analysis of ten different machine learning is carried out using the DLR Urban Traffic dataset. The experimental results indicate that the Random Forest classifier surpasses the performance of other classifiers. The random forest classifier achieved an overall accuracy of 99.30% in predicting individuals' travel modes. The evaluation of the proposed system is carried out using different metrics such as precision, recall, accuracy, and AUC curve. This demonstrates that the proposed research can be effectively applied in real-time scenarios to analyze travel behavior and recommend suitable travel modes for individuals based on specific constraints with high precision.