Understanding how commuters choose their travel modes is essential for fostering sustainable mobility and reducing reliance on private vehicles. This study employs Multinomial Logit (MNL) and Neural Network (NN) models on survey data collected from employees in Rome, Italy, ensuring a fair comparison by applying identical preprocessing and evaluation metrics. While the NN model achieves slightly higher accuracy, statistical analysis indicates that the difference is not significant. The elasticity analysis highlights key factors shaping commuting choices, offering interpretable insights into travel behavior. These results demonstrate that, despite its lower accuracy, MNL provides strong predictive capabilities while preserving interpretability. This underscores the continued importance of traditional econometric models in transportation research, particularly for policymaking, where explainability is crucial.